Content uploaded by Matthew Ponting
Author content
All content in this area was uploaded by Matthew Ponting on Nov 14, 2018
Content may be subject to copyright.
K.
BUTCHER and
M.
PONTING
ROME AND THE EAST. PRODUCTION
OF
ROMAN PROVINCIAL
SILVER COINAGE FOR CAESAREA IN CAPPADOCIA UNDER
VESPASIAN, AD
69-79
Summary
During the first three centuries AD several eastern provinces
of
the Roman empire produced silver coinages
of
specijically local denominations
and types. It has been claimed that at certain periods the mint
of
Rome was
employed to strike some
of
these coinages, rather than the provincial mints
to which the coins are traditionally attributed. This claim is based mainly on
stylistic observations, but because style is regarded as subjective, some other
form
of
evidence to support these observations is desirable. In this paper three
types
of
coin are submitted to metallurgical analysis: silver denarii struck at
Rome; Roman-style provincial silver coinage; and ‘local’ style provincial silver
coinage. The aim is to discover whether the Rome-style and ‘local’ style coins
have different trace element projiles, and whether these might indicate different
ore sources or reJining techniques. The results are then compared with Roman
denarii to see whether there are any similarities between the denarii and the
Rome-style provincial silver coins.
INTRODUCTION
Various groups of Roman silver coins,
issued for the eastern provinces and composed
of denominations which did not circulate in the
central or western Roman empire, exhibit
stylistic and technical characteristics which
suggest that they are products of the mint of
Rome. This is acknowledged in various publi-
cations (e.g. Butcher
1988, 36-7),
but
nowhere are the phenomenon or its parameters
explicitly defined. Until the present study the
observation has rested solely on the basis of
style and minor technical points. The period
when this phenomenon was most intense seems
to have been between the reigns of Vespasian
OXFORD
JOURNAL
OF
ARCHAEOLOGY
14(
I)
1995
and
238
Main
Street. Cambridge.
MA
02142.
USA.
I
Ba\il Blackwell Ltd
1995.
108
Cowley Road. Oxford OX4
IJF.
UK
(AD
69-79)
and Trajan (AD
98-
117),
when
several important eastern silver coinages seem
to be very Roman in style: the
cistophori
of
Asia (modern western Turkey); the drachm
coinage of Lycia (southern Turkey); certain
tetradrachms of Tarsus (a city in southern
Turkey); the coinage of Caesarea in Cap-
padocia (Kayseri in central Anatolia); the
coinage of Antioch in Syria (Antakya, also
in
Turkey); the coinage of the province of Arabia,
often thought to have been struck at Bostra
(modem Busra Eski-Sham, Syria); and another
series of coins, only recently identified as
issues for the province of Cyrenaica (covering
parts of modem Egypt and Libya). These
63
ROMAN PROVINCIAL SILVER COINAGE
provinces are thought to have possessed their
own currency systems which were not directly
compatible with the unified system used by the
central and western provinces of the empire;
many were probably the legacy of pre-Roman
currency systems which the Romans saw no
advantage in changing (Butcher, forthcoming).
The provinces normally produced their own
silver coinages at a central provincial mint.
Caesarea struck a copious silver coinage for
the Anatolian province of Cappadocia, and by
the late first century AD Antiochene silver
tetradrachms circulated all over the Levant.
The identification of a Roman style among
many of these coinages between AD 69 and
1
17 is not the end of the story. Some of these
mints appear to have produced silver coins
which exhibit local styles and techniques of
manufacture
contemporaneously
with the
Rome-style issues. This is particularly apparent
at Caesarea in Cappadocia. There was there-
fore a regular local supply, with the possibility
that this supply was supplemented by coins
struck at Rome and sent to the province.
However, the possibility that Roman mint-
workers, or Roman dies, were sent to the
provinces in question rather than the finished
product could not be ruled out. Fortunately in
some cases there is some evidence to discount
the idea that the dies were sent from Rome and
used by local mint-workers. These are con-
sistent differences in the technique of manu-
facture between Rome-style issues and certain
local style issues. The dies were positioned
differently when striking Rome-style coins,
so
that the image on the reverse is always inverted
with respect to the obverse (the reverse at a
'6
o'clock' or 180"axis). This was normal
practice at the mint of Rome between the reigns
of Vespasian and Trajan. In a mint such as
Caesarea in Cappadocia the practice was to
strike coins with the dies aligned with one
another (the reverse at '12 o'clock' or
0").
The
Rome-style coins of Caesarea in Cappadocia
always have a 6 o'clock axis, and local style
12 o'clock. This in itself suggests that it was
not the dies which were being sent from Rome
and used by local mint-workers, unless they
broke with tradition when using those dies.
However, the possibility that Roman mint-
workers went to Caesarea remained. In the
case of certain other local coinages, the
distinction between Rome style and local style
is less apparent. The local issues were struck
with the dies at the 6 o'clock position, like the
coinage of Rome, and no distinction can be
made except for style.
The aim of this paper is to examine the
elemental profiles of three
types
of silver coins:
those which we know were struck at Rome
(the
denarii);
certain Rome-style eastern coins; and
certain local style eastern coins; and to look
for similarities and anomalies between the
three. A similarity between Rome-style coins
and denarii, and a dissimilarity between these
two types of coin and the local style pieces,
would indicate that the Roman-style pieces
were struck at Rome and sent out to the eastern
provinces.
No
detailed elemental analyses of
the silver coinages of this period exist (bronze
coins were examined by Carradice and Cowell,
1987). This report gives details of the first full
analyses of these coins ever performed. The
Department of Coins and Medals at the British
Museum has very generously granted us access
to its collections for sampling and we would
stress that without the enthusiastic support of
its staff this project would have been quite
impossible.
THE COINS EXAMINED
To investigate the possibility of Rome
producing coins for the eastern provinces we
chose a particular group of provincial silver
coins which exhibited both Roman and 'local'
styles: coins of Caesarea in Cappadocia struck
under Vespasian (AD 69-79). These were
OXFORDJOURNALOF ARCHAEOLOGY
64
0
Basil
Blackwell
Ltd.
1995
K.
BUTCHER
and
M.
PONTING
issued in three denominations.
1)
a large, thick
coin of about
7
grammes, normally called a
didrachm or two-drachm unit;
2)
a smaller unit
of about half the weight (a drachm); and
3)
a fractional piece, thought to be a half-drachm
or hemidrachm. Because of their small size,
fragility and thin flans, no hemidrachms were
sampled. The didrachms and drachms were
issued in both Rome style and local style
(hemidrachms only occur in local style). Their
designs are virtually identical, as are their
weights. Coins were issued bearing the names
and portraits of Vespasian’s sons Titus and
Domitian as well as the emperor himself. The
die-axes of the Rome-style pieces are
6
o’clock, the local
12
o’clock. Many of the
coins bear no date of issue, but some of them
are dated by regnal years of Vespasian: six,
seven, and, most commonly, nine. Since
undated coins share dies with dated pieces, it
seems likely that the undated pieces were
produced at the same time as the dated coins.
The exact relationship of the Caesarean year
to the calendar year is uncertain, but year six
must fall between AD
73-5;
year seven
between AD
74-6,
and the ninth regnal year
between AD
76-78.
The whole period of
production may be assumed to belong to the
period AD
73-78,
perhaps with some gaps,
and perhaps extending either side of these
limits (Walker
1976, 128
mentions hemi-
drachms of Titus as Augustus, which would
extend the issue down to AD
79
at least).
The British Museum collection does not con-
tain specimens of every type of didrachm and
drachm issued at Caesarea under Vespasian,
and as a result we do not have analyses for
the full range of issues. We do, however, have
a representative sample of the principal types,
taken from nineteen coins, and as our results
show, the coins form discrete, definable
groups.
The specimens analysed are briefly des-
cribed in the following list. The order in which
they occur, and in which they were sampled,
is simply the order in which they are kept in
the British Museum’s tray of Flavian
Caesarean material. The reverse types are
more fully described in Sydenham
(1933).
Vespasian
1.
Didrachm.
Local
style. Undated. Reverse:
Nike standing right. BM
1938 10-7-73.
2.
Didrachm.
Local
style. Undated. Reverse:
Nike standing right. BM
1920 2-22-1.
3.
Didrachm. Local style. Undated. Reverse:
Nike standing right. BM
1931 6-1-43.
4.
Didrachm.
Local
style. Undated. Reverse:
Nike standing right. BMC
16.
5.
Didrachm. Rome style. Undated. Reverse:
Mount Argaeus. BM
1927 12-2-2.
6.
Didrachm. Rome style. Undated. Reverse:
Mount Argaeus. BM
1938 10-7-74.
7.
Drachm. Rome style. Undated. Reverse:
Nike standing right. BM
1931 6-1-54.
Vespasian and Titus
8.
Didrachm.
Local
style. Undated. Reverse:
Head of Titus. BM
1979 1-1-1108.
9.
Didrachm.
Local
style. Undated. Reverse:
Head
of
Titus. BM
1931 6-1-42.
10.
Didrachm. Rome style. Undated. Reverse:
Head of Titus. BMC
19.
11.
Didrachm.
Local
style. Undated. Reverse:
Head of Titus. BM
1921 10-4-1.
12.
Didrachm. Local style. Year
9.
Reverse:
Titus standing. BMC
20.
13.
Didrachm. Rome style. Year
9.
Reverse:
Titus standing. BM
1938 10-7-76.
Vespasian and Domitian
14.
Didrachm. Rome style. Year
9.
Reverse:
Domitian standing. BM
1979 1-1-1109.
15.
Didrachm. Local style. Year
9.
Reverse:
Domitian standing. BM
1927 11-7-1.
16.
Didrachm. Local style. Year
9.
Reverse:
Domitian standing. BM
1931 6-1-44.
OXFORDJOURNALOF ARCHAEOLOGY
0
Basil Blackwell
Ltd.
1995
65
ROMAN PROVINCIAL SILVER COINAGE
17.
Didrachm. Rome style. Year
9.
Reverse:
Domitian standing. BM
1938 10-7-77.
18.
Didrachm. Rome style. Year
9.
Reverse:
Domitian standing. BM
1927 12-2-3.
19.
Didrachm. Local style. Year
9.
Reverse:
Domitian standing. BM
1927 3-5-105
1.
Two further specimens of the Rome-style
coinage were generously placed at our disposal
by a private collector. These are interesting
in that they share obverse and reverse dies,
and both have been analysed previously using
a different technique (X-Ray Fluorescence) to
the one employed here (Atomic Absorption
Spectrometry).
Vespasian
20.
Drachm. Rome style. Year
6.
Reverse:
2
1.
Drachm. Rome style. Same dies as no
20.
Mount Argaeus. Private coll.
Private coll.
The results of the analyses of these
Caesarean coins were then compared with
Roman denarii of the period AD
76-78.
The
difficulties of deciding which denarii to sample
were compounded by the presence of a large
number of ancient plated forgeries in the
museum collection. In fact there were many
fewer suitable specimens than anticipated, but
the number was sufficient for our purposes.
The British Museum staff granted us per-
mission to sample some additional denarii from
recent hoards. Unfortunately these coins had
been aggressively cleaned, and after analy sing
them we found that their internal compositions
had been seriously altered by the cleaning
process, which had removed a high proportion
of their copper content. It was therefore
decided not to include them in the present
study.
The denarii analysed are briefly described
in the following list. References are to
Mattingly’s British Museum Catalogue
(1930).
1.
Obverse: Head of Vespasian facing right.
Reverse: COS VIII. Mars holding spear
and trophy. BMC
200.
2.
As previous. BMC
200a.
3.
As previous. BMC
201.
4.
Obverse: Head of Vespasian facing right.
Reverse: COS VIII. pair of oxen. BMC
206.
5.
As previous. BMC
207.
6.
As previous. BMC
208.
7.
Obverse: Head of Vespasian facing right.
Reverse: COS VIII. Prow. BMC
210.
8.
As previous, but head of Vespasian facing
left. BMC
21
1.
9.
Obverse: Head of Titus facing right.
Reverse: COS VI. As nos.
1-3.
BMC
222.
10.
Obverse: Head of Titus facing right.
Reverse: COS VI. As no
4.
BMC
225.
ANALYTICAL TECHNIQUE
Sampling problems
When a predominantly binary allow consists
of two metals one of which is significantly
more electro-negative than the other, the more
electro-negative metal will be preferentially
corroded out. This will lead to an apparent
increase in the observed concentration of the
more electro-positive constituent within the
area susceptible to corrosion processes. This
phenomenon is known as surface enrichment
because it is at the surface of a metal object
that this will occur. However, the depth of the
area affected can often vary considerably
depending on both the nature of the sur-
rounding environment and the nature of the
alloy. This natural process can be further
complicated by purposeful enrichment
(blanching) of the coin blanks prior to striking
resulting in a double enriched zone. The coins
examined in this paper are all silver-copper
alloys and the effects of surface enrichment
OXFORDJOURNALOF ARCHAEOLOGY
66
0
Basil
Blackwell
Ltd.
1995
K.
BUTCHER and
M.
PONTING
can cause serious problems for the accurate
analyses of this material.
In an article published in
1964,
Condamin
and Picon examined the preferential corrosion
of copper in denarii of Septimius Severus
(Condamin and Picon
1964)
and concluded that
the surface silver content of debased silver-
alloy coins will be markedly higher than the
silver content of the alloy within the coin. It
was also demonstrated that the metal within
the coin was more truly is representative of
the original alloy. In particular this article
demonstrated the problems of Neutron Acti-
vation Analysis (NAA) as applied to silver-
alloy coins. This technique accurately analyses
the whole coin as it is today, including the
surface enriched areas. A total NAA analysis
gave a figure
of
53
%
silver compared with an
internal silver content of
45%
for a Severan
denarius (Condamin and Picon
1964, 104).
Furthermore, it was also shown that the
enriched area can often extend some way into
the coin. Despite this, and other work (Cope
1973)
demonstrating the serious problems
associated with such analyses, many studies
have been conducted claiming to obtain
numismatically meaningful results from
surface analyses (Walker
1976).
Carter
(1973)
tried to get around this problem for X-Ray
Fluorescence analysis (XRF) by air-abrading
the area to be analysed removing a total
estimated
60
pm of surface metal. This was
the amount of metal removed in order to
achieve a repeatable reading (three repeats
starting at
20
pm, ending at about
60
pm), the
assumption being that this representative of the
main bulk of the coin. However, it appears that
this is misleading. There is a notable
discrepancy between the silver values for
Severan denarii issued after AD
193
as
reported by Condamin and Picon
(1964)
and
as reported by Carter
(1973).
Carter gives us
an average figure of
58.4%
f
0.5%
(normalized) based on ten analyses
(1973,70).
This is a discrepancy of over
13%.
Yet it is
Carter’s method which has been repeatedly
used for numismatic research. Indeed, it is this
method which was used by Walker in his
magnum
opus
‘The metrology
of
the Roman
silver coinage’
(1976-78)
and gave a mean
silver value of
57.59%
for the same issues as
used by Carter.
’
The apparent discrepancy
between Carter’s and Walker’s results and the
results of Condamin and Picon would suggest
a fundamental flaw in the application of XRF
to coin analyses and convinced us of the need
for a more rigorous methodology employing
Atomic Absorption Spectrometry (AAS). The
results of these analyses confirm this and the
significance of this method is discussed below.
Sampling strategy
The aim of the sampling strategy adopted
for this project was to provide an adequately
large and representative sample of the alloy
from which the coin had been produced. The
effects of corrosion together with a likelihood
of an intentional enrichment of silver at the
coin’s surface (blanching) means that a sample
needs to be taken from well beneath the coin’s
surface. This was achieved by drilling two
approximately equidistant holes into the cylin-
drical edge of each coin analysed and collecting
the turnings. The first millimetre or
so
of
drillings was always discarded to try to ensure
that there was little
or
no contamination from
the
surface layers: the coins usually being thick
enough to ensure that little or no enriched
material was picked up from either face. Once
taken the turnings were amalgamated to form
the
10-15
mg sample required for the
analysis.
Method
Atomic Absorption Spectrometry was
chosen as the most appropriate technique to
OXFORDJOURNALOF ARCHAEOLOGY
0
Basil Blackwell
Ltd.
1995
67
ROMAN PROVINCIAL SILVER COINAGE
TABLE
I
PERCENTAGES
OF
METALS PRESENT IN FLAVIAN CAESAREAN COINAGE AND CONTEMPORARY DENARII OF ROME
smy
kf
WI
c.u
nm
mwr
Cappr
Aa111my
Ir.d
lrm
ml1
Blirrb Z1.c
RLd
ToIl
vupaa.
114381D7-bCd
1
lp202-22-bCcll
J
1931 614bcd
4BMClb
bCd
5
1927 12-2-Rane
6
143810-7-Rane
7
1931
bl-5Ranod
zoprime
Rane-d
lIpivr*
Ranod
vnpamaad
nim
a
1979
1-1-1
bcd
9
1931
614bCd
II
1921
104bcd
11BMCm
bcd
108Mc19
Rane
lpyl10-7-Rane
vupsamatd
mnmn
15192711-7-bcd
191927Slbcd
14
1979 1-1-1
Rane
1’1
1438 W7-m
16
19-31 bldbcd
I*
1927 12-2-Rane
DL~III
d
m.x7a
leMcl00
Ran0
1BMcZoRARane
JBMC201
Rome
4BMc2ob
Ran0
5BMC207
Ran0
6BMCXr)
Ran0
7BMc210
Rane
IBMC211
Rome
9BMQp
Ran0
10-
Rane
0.3
0.28
0.32
0.22
0.24
0.2
0.24
0.21
0.31
0.34
0.29
0.33
0.26
0.23
0.18
0.3
0.27
0.27
0.15
0.17
0.17
0.30
0.38
0.43
0.36
0.32
0.42
0.47
05
0.28
0.35
0.3
0.23
0.27
0.25
0.18
0.17
0.17
0.1
1
0.09
0.18
0.17
0.24
0.2
0.19
0.24
0.25
0.24
0.15
0.22
0.17
0.18
0.10
0.28
0.2
0.lb
0.24
0.2
0.27
0.24
0.22
0.1
7
4b.a
47.31
475
47s
4b.
74
4a53
4b.b
47.m
47.87
011
78
47.46
46.W
47.14
47.07
4b.5
4611
47.51
47.30
47.93
47.47
am
77.w
7b.W
7404
77.19
78.30
80.2
8b.71
79.22
78.4
7a4
09.42
09.w
50.17
w.60
49.75
09.51
09.s
m
49.05
09.02
49.47
50.12
50.57
50.87
09.74
50.76
09.M
50.33
50.W
09.M
mm
19.91
m56
23.71
19.71
1
8.69
17.01
la04
la47
19.2
18.97
7.03
0.03
0.06
0.14
0.18
0.04
0.01
0.02
0.07
0.17
0.24
0.00
0.15
0.07
0.05
0.07
0.09
0.01
0.04
0.03
0.01
0.01
0.01
0.02
0.01
0.01
0.01
0.04
0.03
0.02
0.01
0.01
0.01
0.17
0.19
0.21
0.21
0.41
0.31
0.89
0.51
0.53
0.28
0.24
0.32
0.17
0.62
0.41
0.19
0.19
0.17
0.80
047
059
0.66
0.74
0.35
0.65
0.71
0.79
1.29
0.38
0.6
0.64
0.09
0.01
0.01
0.01
0.01
0.07
0.03
0.09
0.05
0.03
0.01
0.01
0.03
0.01
0.1
0.11
0.M
0.04
0.04
0.M
0.09
0.03
0.05
0.07
0.03
0.03
0.03
0.02
0.02
0
0.02
0.01
0.01
0.01
0.01
0.01
0.01
0
0.01
0.01
0
0
0
0.01
0.01
0
0.01
0.01
0
0.01
0
0.01
0.01
0
0.W
0.09
0.03
0.03
0.04
0.01
0.03
0.03
0.03
0.03
0.04
0.02
0.M
0.04
0.05
0.01
0.01
0.03
0.04
0.05
0.04
0.01
0.05
0.06
0.03
0.01
0.04
0.M
0.05
0.06
0.03
0.04
0.17
0.2b
0.08
0.14
0.14
0.06
0.13
0.07
0.07
0.09
0.06
0.02
0.01
0.01
0.01
0
0
0.01
0.01
0.04
0.01
0.02
0
0
0
0
0
0
0
0
0
0
0.02
0.01
0.06
0.01
0.02
0
0
0
0
0.01
0.03
0.03
0.03
0.03
0.04
0.01
0.01
0.04
0.03
0.03
0.03
0.04
0.03
0.01
0.03
0.01
0.03
0.03
0.03
0.01
0.03
0.03
0.03
0.04
0.01
0.01
0.01
0.01
0.02
0.02
0.02
0.01
0.01
9b.0
97.85
97.b7
97.84
W.1b
98.66
98.76
98.53
?am
98.89
97.7b
9az
Pam
W.18
97.25
97.8
98.45
97.37
98.92
98.50
pa17
W35
98.98
98.W
98.27
98.75
98.99
98.93
98.82
98.b7
W.8b
qa&
obtain the accurate bulk analyses required for
this project. This technique requires that all
the samples are in solution. The analyses of
silver alloys present a problem in that in order
to get all the elements present into solution it
is necessary to use a mixture of nitric
(HN03) and hydrochloric (HCl) acids called
aqua regia. Unfortunately silver complexes
with the chlorine
in
the HCl to form a sparingly
soluble precipitate
of
silver chloride. The only
way to avoid this is to use a very high con-
centration of acid
(50%)
which is undesirable
as the viscosity of the solution can interfere
with the analysis. Various ways around this
problem have been suggested (Hughes
et
al.
1976, 27).
A
variation of one of these was
applied here. This involves using two
solutions; the first employing the
50%
aqua
regia on a very small (2-3 mg) sample weight
and only for determinations of tin and gold (the
elements not dissolvable in HN03), the other
using the bulk of the sample
(=
10
mg)
dissolved in HN03. This second solution is
used
for the determination of all other elements
investigated.
The elements analysed for were silver,
copper, lead, tin, zinc, gold, iron, nickel,
bismuth, cobalt and antimony. Normal
operating procedures were used, with all
standard solutions matched to the matrix and
acid concentrations of the samples. The
precision of the technique
is
f
1-2%
for
OXFORDJOURNALOF ARCHAEOLOGY
68
0
Basil Blackwell
Ltd.
1995
K.
BUTCHER
and
M.
PONTING
major components,
*5%
for minor com-
ponents and
*20%
for trace components. The
results are listed in Table
1
(p.
68).
In order
to provide an independent check on the silver
and copper values a sample of Sterling silver
was run alongside the analysis. The results for
this were well within expected experimental
error.
In addition to AAS analyses a sample of
drillings was analysed by
ED-XRF
to confirm
that no important elements had been left out
of the analysis.
RESULTS
The Caesarean coinage
The silver contents of the coins analysed
were considerably lower than expected. The
mean silver content for the Caesarean issues
(both Rome style and local) was
47.15%
with
a standard deviation of
0.84
and the copper
50.07%
with a standard deviation of
0.51.
Walker’s analyses of the same coinage give
a mean silver content of
62.91
%
and a standard
deviation of
7.62.
No
significant differences
were found in the silver levels between the
two
stylistic groups (means of
47.27%
for local
issues and
46.99%
for Rome-style issues). The
low standard deviation is interesting, sug-
gesting a very accurately controlled alloying
process, whether at the mint of Rome or at a
local mint. This indicates that Walker’s high
standard deviation was solely a product of
differences
in
the effectiveness of the blanching
process and degree of corrosion suffered. The
figures clearly point to a
5050
alloy being
aimed at, which was
very
rigidly adhered to
by the producers of both Rome-style and local
style coins.
As
an independent check on the
AAS determinations drillings from one of the
coins were analysed by Electron-probe micro-
analysis (EPMA). Ten separate turnings were
analysed from coin
No.
18.
These gave a mean
of
47.72%
which agrees favourably with the
47.47%
given by AAS.
However, although the silver and copper
ratios are virtually identical for both Rome-
style and local style issues, our analyses also
show that the trace elements divide the two
groups neatly. The concentrations of gold, lead
and bismuth are particularly useful in this.
Gold and bismuth are probably the two
elements which are least affected by the silver
smelting and refining processes. Consequently
the marked separation apparent when the gold
and bismuth values are plotted is likely to
reflect different metal/ore sources (Figs.
1
and
2)
*
The majority of silver used in antiquity
would have been extracted from argentiferous
lead ore (galena) using the process of cupel-
lation. This method involes first smelting the
ore to produce a silver-rich lead, which was
then oxidized to remove most of the lead as
litharge (lead oxide) and concentrating the
silver in the remaining lead. The silver would
then be extracted in ‘cupeles’ (small crucibles)
containing bone ash. This removes most of the
remaining lead, but some inevitably remains,
providing us with an indication of the effective-
ness of the procedure employed. Experimental
work reported by Tylecote
(1986,61)
indicated
that the lead content of silver produced by this
process is unlikely to fall much below
1
%.
However, the figures from these analyses
suggest that, in this case, lower levels were
indeed achieved. Figure
2
shows the lead
values plotted against the gold values and
clearly shows that the local and Rome issues
are
also
distinguished by their lead levels. This
points to a level of distinction between the
groups other than the ore/metal source: that
of technology used to produce the alloy. The
‘Rome’ style issues clearly have a significantly
higher lead content than the ‘local’ style issues.
Furthermore, the ‘local’ style issues form a
much more discrete group which
may
indicate
OXFORDJOURNALOF ARCHAEOLOGY
0
Basil
Blackwell
Ltd.
1995
69
ROMAN PROVINCIAL SILVER COINAGE
0.07
0.M
a05
P
g0.04
d
E
0.m
if
0.m
0.01
-
--
-.
--
--
-.
--
0
0.9
ae
0.7
0.6
0.5
*
a
a
0.4
0.3
0.2
0.1
f
04
0
0
0
-
0
0
..
.-
--
0
0
0
0
0
-
.-
.-
0
0 0
0
ooOo
Po
-
..
I
0
0
000
0
0
0
O
El
0
0
0
0
0
0
I
0.00
0.05
0.10 0.15
0.20
0.25
0.30
0.35
Gdd(Wt.%)
Figure
1
CAESAREA: gold content
vs
bismuth
content
a smaller production unit. This
is
what may
be expected for a local provincial mint
producing comparatively small numbers of
coins. The coins analysed were all issued over
a period of only a few years and
so
are discrete
chronologically. It is relatively unsurprising,
therefore, that the local products exhibit a
greater compositional similarity using an
element which reflects the refining processes
used than those coins thought to have been
produced at the mint of Rome.
The denarii and their relationship to the
Rome-style Caesarean coinage
We then sampled ten Roman denarii of AD
76-78. The silver contents differed markedly
from those published by Walker (1976) as we
had expected. Despite slight contamination
from the enriched zone (because of the thinness
of some of the coins sampled) a mean value
of 78.71
%
was obtained. This contrasts
significantly with Walker's means of
90.86%
OXFORDJOURNALOF ARCHAEOLOGY
0
Basil Blackwell
Ltd.
1995
70
K.
BUTCHER and
M.
PONTING
250.00
m'ml
.8
8
m8
80
0
8
8
8
0
0
ooo
0
00
O
0
0
0 0
0
0.00
4
4
o.m
50.00
1m.m 1m.m
moo
250.00
300.00
350.00
ACCAu
Figure
3
CAESAREA AND ROME. Ratio plot
of
silver:gold and silver-lead values
(13 coins of AD 76) and 89.77% (28 coins
of
AD 77-78) (Walker 1976,90-1). The other
results obtained show a good correlation
between the trace element profiles of the
Rome-style coins and the denarii, and a
discrepancy between these two groups and the
local style coinage (Fig. 3). Because of the
differences in silver contents between the
denarii and the Caesarean issues it was decided
to use ratios rather than concentration levels.
The ratios of gold to silver and lead to silver
should remain more or less constant between
coins made of silver from a similar source.
This was found to be the case. Figure 3
illustrates the plot of the two ratios and clearly
shows the denarii and Rome-style Caesarean
issues to be of
very
similar composition. Two
of the denarii are, however, attributed to the
local field on the basis
of
their ratios. This is
interesting given that we are not dealing with
outliers but with
an
unexpected attribution. The
provenance of one, no. 3, is unusual: it was
found in India, but
in
its type content there is
nothing which distinguishes
it
from nos. 1,2,
or 9, which are thought to be coins of the same
issue. The provenance of the other anomalous
coin, no. 8, is not known; again, its type is
the same as no. 7, which fits well in the Rome
field, although no.
8
is distinguished by having
a left-facing bust of Vespasian on the obverse,
which might indicate that
it
is part of a separate
issue
from no. 7. It
is
possible that these two
denarii were indeed struck from metal of
eastern provenance; we propose to examine
some more specimens of these two coin types
in other collections to
see
whether the anomaly
extends to other examples of the issues.
To investigate the differences between the
Rome-style and local style issues further,
principal components analysis (PCA) was
used. This is a multivariate statistical technique
which is used to investigate structure in a data
set consisting of many variables (Baxter
1993a). Principal components analysis
attempts to visually plot all the cases in two-
dimensional space, and to look for groupings
within the plot. To do this a new set of
variables are computed which are linear com-
binations of the original variables. Statisticians
tell us that there is then a good likelihood that
the first two or three of these will reveal the
most important features within the data set
(Baxter 1993b, 1). These new variables
(principal components) are therefore plotted
against each other
in
order to view the structure
revealed. The first axis is the first principal
OXFORDJOURNALOF ARCHAEOLOGY
GI
Basil Blackwell
Ltd.
1995 71
ROMAN PROVINCIAL SILVER COINAGE
1-
go-
-1
-2
-3
-4
.
.
0
.
0
0
0
0
Rome drachms
00
0
-
0 0
Rome didrachms
0
Local didrachms
0
-
0
0
I I
I
I
I
Figure
4
CAESAREA: PCA plot
of
rank order transformed data
component (pcl), the second axis is the second
principal component (pc2), and
so
on, there
being as many axes as variables. However,
with this techique, the assumption is made that
the greatest proportion of the variability is in
the first two components. There are several
types of PCA which can be used, all different
by virtue of the transformation process carried
out on the data prior to the analysis proper.
As
a
general rule it should be stressed that no
single method produces the ‘correct’ results.
The approach adopted here has been to
employ a selection of carefully chosen tech-
niques with which to investigate structure
within the data set. Four types of data
transformation were used: natural log, stan-
dardized raw data, rank-order transformation
and Aitchison’s log-linear transformation
(see
Wright
1991, 37-9).
In all cases PCA of the
transformed data show a clear separation of
the local and Rome-style issues. The log-
transformation is used because it tends to
produce variables with approximately equal
weight (variances). This is important with data
like this, where the analysis would otherwise
become dominated by the largest variables
present (the silver and copper) (Aitchison
1986).
Wright suggests the use of the rank-
order transformation (Wright
1991,
38)
as a
simple but effective method of satisfying
assumptions of linearity in a data set. This
method is not widely used amongst archae-
ological scientists but is regarded by Wright
as a more useful option than Aitchison’s log-
linear transformation (Wright
1991,
38),
especially as it is not overly sensitive to small
values.
Generally speaking each method of trans-
formation produced a plot which clearly
separated the Rome-style and local style coins
(Fig.
4).
Closer investigation of the variable
loadings suggest that, apart from the lead, gold
and bismuth, iron, tin and antimony are
significant contributors to the groupings.
Nickel and cobalt may also play a part;
however the role played by these metals is not
substantiated
by
all
the methods of PCA
employed and
so
may be a function of the
transformation processes alone. Going back to
the raw data we see that the local issues
generally appear to have higher levels of tin
(0.23%
against
0.19%)
and antimony
(0.10%
against
0.03%)
than the Rome issues but lower
levels of iron (0.02% against
0.07%).
These
OXFORDJOURNALOF ARCHAEOLOGY
72
0
Basil
Blackwell
Ltd.
1995
K.
BUTCHER
and
M.
PONTING
'
0.12
0.1
0.08
-
.\q
L
OM
0
-
0.04
om
-.
01
-
0
..
0
00
-.
0
0
0
.-
-.
0
0.
0
0
00
0
0
I
differences are indeed very small, but their
significance is clearly demonstrated in the plot
of antimony against iron values in Fig.
5.
In the same way that we have suggested that
the lead, gold and bismuth levels found in these
coins are related to the silver, its sources and
refining processes,
so
too are the antimony and
carefully removed by fire and melted out
of it becomes bar copper
. .
.
At Capua
it
is smelted in a fire of wood, not of charcoal,
and then poured into cold water and cleaned
in a sieve made
of
oak
.
. .
this process of
smelting is repeated several times
.
.
.
(Pliny,
HN
34, 94-95).
iron related to the copper. The differences in
iron content between the two groups is
interesting. Iron is an impurity present in
smelted copper and comes from the use of iron
oxide as a fluxing agent. After its initial
smelting a crude 'black copper' is produced
containing several percent of iron. This
material is very brittle and unsuitable for most
uses. Furthermore, the Romans were well
aware of this fact as we see from this section
in
Pliny
's
Natural History:
Bar copper also is produced in other mines,
and likewise fused copper. The difference
between them is that the latter can only be
fused, as it breaks under the hammer,
whereas bar copper, otherwise called ductile
copper, is malleable, which is the case with
all Cyprus copper. But also in the other
mines, this difference of bar copper from
fused copper is produced by treatment; for
all copper after impurities have been rather
A simple refining process is therefore
sufficient to remove most of the iron and
improve the working properties and any subse-
quent re-melting will further reduce the iron
content. Thus the iron level in copper can be
used as an index of the level of refining. This
suggests that the local mint was using either
better refined
or
more frequently remelted
copper than the mint of Rome.
A
similar
phenomenon has been noted
in
the composition
of copper-alloy coins of the fourth century and
their copies (Ponting
1994, 174-6).
There it
is suggested that either the official mints were
using freshly smelted (and refined) copper
whereas the copies were produced from re-
melted scrap
or
that the small scale of the
industry producing
the
copies enabled a higher
level of refining of freshly smelted copper. A
similar model can be suggested here, with the
local mint relying on re-cycled old copper-
alloy coins refined to remove any valuable tin
OXFORDJOURNALOF ARCHAEOLOGY
0
Basil
Blackwell
Ltd.
1995
73
ROMAN PROVINCIAL SILVER COINAGE
whereas the mint of Rome was using freshly
smelted copper refined only to the level
required to make-it workable. This model is
further supported by a generally higher tin
trace in the local issues
(0.23%
against
0.19%).
CONCLUSIONS
It is readily apparent from the analyses of
the Caesarean coins that a
5050
silverkopper
alloy was rigidly adhered to at this period for
both drachms and didrachms regardless of
mint. In practice, an exact balance of
5050
for silver and copper was impossible for the
ancient mints to achieve because of the
impurities present in the metals, but at such
low levels the Romans were almost certainly
unaware of the presence of these impurities.
The trace element profiles demonstrate that the
copper and silver
used
to make this alloy differ
significantly between the two numismatically
defined issues by virtue of their impurity
levels. Furthermore, the ratios of trace
elements associated with the refining of the
silver are basically the same for denarii issued
at Rome and Caesarean coins of the Rome
style. This would suggest a common metal
source and/or refining procedure for both and
therefore supports the view that the Rome-style
Caesarean coins were indeed struck at Rome.
Both of the main constituents of the alloys of
the Caesarean coins, silver and copper, contain
trace elements strongly associated with the
technology
used
in smelting and refining them.
The fact that trace elements associated with
both groups, Rome style and local style, differ
significantly from one another also supports
the view that the origins of the silver and
copper used
in
the issues are different. In
addition, the differences in the trace element
profiles also allow suggestions to be made
concerning the differences in the technologies
used to produce the alloys. These
will
obviously reflect mint practice and allow a
glimpse of the attitudes and standards operating
in the mints at the time.
Acknowledgements
No project of this kind can be undertaken without
appropriate funding, access to coins for sampling, and
equipment necessary to perform the analyses. We are
particularly indebted to the members of the relevant
committees who selected this project for funding. None
of the research could have been contemplated without
sufficient grant support, for which we are extremely
grateful. The Leverhulme Trust generously provided a
Fellowship for Dr. Kevin Butcher
to
undertake a
programme of analyses of Roman provincial silver
coinages from AD
69
to
117,
and the British Academy
kindly awarded the project a grant from their fund for
applied science in archaeology to support part-time
research assistance by Dr. Matthew Ponting. Silver and
copper standards used in the analyses were provided free
of charge by Johnson Matthey
PIC.
Special thanks are
reserved for the staff of the institutions in which the work
has been undertaken: in particular Dr. Andrew Burnett
and Dr. Roger Bland of the British Museum’s Department
of Coins and Medals and Mr. Michael Cowell of the
British Museum Research Laboratory, not only for access
to material but also for their support, patience and for
many helpful discussions; and to Dr. John Merkel and
Dr. Dafydd Griffiths of the Institute of Archaeology,
London, for access to equipment and for advice.
Professor Theodore Buttrey and Dr. Richard Reece have
also provided much help and advice on the numismatic
front. Without the support and encouragement
of
all it
would have been quite impossible to undertake the work
presented above.
(KB)
Faculty
of
Classics,
Sidgwick Avenue,
Cambridge
CB3 9DA
(MP)
British
Museum
Research Laboratory
Great Russell Street,
London
WClB 3DG
74
OXFORDJOURNALOF ARCHAEOLOGY
0
Basil
Blackwell
Ltd.
1995
K.
BUTCHER and M. PONTING
APPENDIX
The silver content and the problem of standards
One aspect which we were not prepared for was the
silver content of the Caesarean and denarius coinages
of
Vespasian. This, as mentioned above, is much lower than
anybody had previously suspected. The extensive series
of analyses performed by D.R. Walker in the
1970s
(Walker
1976; 1977; 1978)
had examined Roman silver
coinages of the same period using ED-XRF. Walker
generally looked only for silver, gold and lead (he
occasionally reported traces of other impurities), and often
gave figures only for silver content, the quantities of the
other two elements frequently being beneath the detection
limits of the technique. He built a complex series
of
arguments
on
his results, which showed the decline of
the silver content in the denarius over time, and a
tendency for eastern silver issues
to
be struck
to
a lower
fineness than the denarius. Nobody has tried to repeat
an analysis
of
the denarius coinage
on
this scale, and his
work has been extremely influential among economic and
monetary historians of the Roman period. His tables
showing the decline of silver in the coinage, and the
relationship between gold and silver coinages, are
standard illustrations in general works
on
Roman coinage.
Walker reached the denarius coinage
of
Vespasian early
in his programme of analyses. He was puzzled by his
results, which showed that the percentages of silver had
a high standard deviation. The
issues
of previous
emperors show a low standard deviation, and his
individual analyses had generally been consistent,
allowing him
to
guess how many parts of base metal had
been added to parts of silver
-
essential for trying to
calculate the standard used by the Roman mint. The
results from Vespasian’s reign onwards were much more
inconsistent, with individual coins showing a wide range
of silver contents. Sometimes two coins, struck from the
same pair of dies, and therefore thought to have been
issued at about the same time, seemed to contain quite
different amounts of silver. Walker concluded that the
mint perhaps had
no
fixed standard or ‘recipe’ when
producing denarii, and that
it
altered the amounts of
copper or silver according to some as yet undefined
pattern. A rough standard could
be
attained by analysing
a large selection of coins of the same date.
When he analysed the Caesarean coinage, Walker
obtained an even wider range of individual readings. He
stated that ‘the Caesarean currency
.
.
.
was struck
on
a very variable standard’ (Walker
1977,
85).
Our results
show that for the reign of Vespasian this was certainly
not true, and
it
is
likely that
it
is untrue for other reigns
as well.
Walker was confounded by the natural processes
of
corrosion, the technology used to produce many of the
coins, and by the analytical technique which he employed.
The standards actually
used
by the Romans for the issues
examined here were very consistent, but with certain
alloys they needed to treat the surfaces of the coins before
issuing them, and any analytical method which examines
the surface only cannot overcome the resulting silver
enrichment. As discussed above, the coinages which we
have examined
so
far contain much less silver than
Walker found, and the amount of variation
is
usually
minimal. Many of the coins which we have sampled are
the same specimens that Walker analysed,
so
there can
be
no
question
of
variation due to different sample
material. He was constrained by a sampling technique
which caused, in his view, the least damage
to
the coin,
by analysing an abraded area
on
the coin’s edge. (In
our
opinion the large abraded area often required for XRF
analyses is much more disfiguring and unsightly than drill
holes, and it is surprising how abrading and removing
surface layers has come to be accepted as a
‘non-
destructive’ technique, whilst drilling small holes into
the edges has not.) Walker’s surface abrading did not
go
very deep, and in very few cases did it penetrate the
surface enriched layers, with the result that his figures
are the result of both the degree
of
enrichment and degree
of penetration of the coin at a given point
on
its edge.
It appears that in the case of the Caesarean issues the
Romans had cast the blanks for the coins from a
prescribed alloy
of
silver and copper which would have
appeared very yellow if left untreated. The discrepancy
between the silver-rich surface layers and the interior is
so
clear that it can be noted visually: the samples removed
from the Caesarean coinage for
our
analyses revealed the
yellow metal of the interior, as did a few (but
unfortunately not many) of Walker’s surface abradings.
In order to preserve the illusion of a ‘fine’ silver coinage
the surfaces
of
the blanks were enriched in silver,
presumably with an acidic solution
to
preferentially
remove the copper from the surface of the alloy.* All
of this would have happened prior to striking. The
resultant ‘honeycomb’ surface
of
purer sitver would have
been compressed and consolidated under the pressure of
the dies. If we compare the silver values obtained by the
different analytical techniques
on
the same coins we can
see a difference in the degree of variation (Fig.
6).
The surface enriched layers of these baser alloys gave
Walker a wide variety of individual values, only the very
lowest
of
which would have approached the
true
silver
content
of
the coinage. Interestingly, he says of the
denarius coinage of Vespasian that whilst ‘a great many
coins’ were
on
the same fineness as coins of earlier
OXFORDJOURNALOF ARCHAEOLOGY
(~1
Basil Blackwell
Ltd.
1995
75
ROMAN PROVINCIAL SILVER COINAGE
8
.
10
04
0
5
10
15
m
Ilsl.No.
Figure
6
Comparison between XRF and AAS analyses of the same coins
emperors, his results produced ‘a “tail” of fineness going
down
to
as low as
c.
78%
fine’ (Walker
1978, 149)
-
our results point to
78%
as the fineness actually used.
Walker’s mean figure
for
this coinage was about
90%
(see above).
The results of our programme of analyses demonstrate
that Walker’s results can only provide
us
with a vague
notion of silver content, and a systematic programme of
analyses of Roman silver coinages will have to be under-
taken if we are to understand the true nature of
debasements and changes to the metal content of the silver
coins of the Roman empire.
NOTES
I.
Condamin and Picon
(1964)
analysed
54
coins of
between AD
194
and
21
1
and gave a figure of between
45%
and
50%
for
36
of the coins. Walker
(1978)
analysed
somewhat more, giving averages of
78.71
%.
65.84%,
66.5%, 61.4%, 57.59%,
58.88%,
55.58%.
55.53%,
57%, 57.67%, 57.16%. 54.75%, 57.07%, 53.21%,
57.63%
and
55.17%
(standard deviations of
4-8)
for
his different issue groups within that period.
2.
A similar technique was used by the Royal Mint to
‘brighten-up’ the low silver content coins issued prior
to the introduction
of
a purely cupro-nickel alloy in
1946.
REFERENCES
AITCHISON,
I.
1986:
The Sratisticaf Analysis
of
Compositional Data
(New York).
BAXTER, M.J.
1993a:
&ploratory Multivariate Analysis
in Archaeology
(Edinburgh).
BAXTER.
M.J.
1993b:
Principal Components Analysis in
Archaeometry
(Nottingham, Trent University Department
of
Mathematics, Statistics and Operational Research).
BUTCHER,
K.
1988:
Roman Provincial Coins:
An
Introduction to the Greek Imperials
(London).
BUTCHER,
K.
forthcoming:
Coinage in Roman Syria.
Northern Syria,
64
BC
-
AD
253
(London, Royal
Numismatic Society).
CARRADICE,
1.
and
COWELL. M.
1987:
The minting of
Roman Imperial bronze coins for circulation in the east:
Vespasian
to
Trajan.
Numismatic Chronicle
147,26-50.
CARTER,
G.F.
1973:
Reproducibility
of
x-ray fluorescence
analysis of Septimius Severus Denarii.
Archaeometry
15,
CONDAMIN.
J.
and
PICON. M.
1964:
The influence of
corrosion and diffusion
on
the percentage of silver in
Roman Denarii.
Archaeometry
7, 98- 105.
COPE, L.H.
1973:
The metallurgical examination of a
debased silver coin of Maximinus Daza
issued
by
Constantine I.
Archaeometry
15,
221-8.
Atomic Absorption techniques in Archaeology.
Archaeometry
18, 19-37.
MATTINGLY. H.
1930:
Coins
of
rhe Roman Empire
in
the
British Museum.
Vol
Il.
Vespasian
-
Domitian
(London,
67-73.
HUGHES, M.J.. COWELL, M.R.
and
CRADWCK. P.T.
1976:
OXFORDJOURNALOF ARCHAEOLOGY
76
0
Basil
Blackwell
Ltd.
1995
K.
BUTCHER and M. PONTING
British Museum).
PONTING.
M.J.
1994:
Folles
and forgeries: an appraisal
of the composition ofRoman copper-alloy coinage of the
mid-third
to
mid-fourth centuries AD from Britain
(Unpublished Ph.D. thesis, Institute of Archaeology,
UCL).
SCHMITT-KORTE. K.
and
COWELL.
M.
1989:
Nabataean
coinage
-
Part
I.
The silver content measured by X-ray
fluorescence analysis.
Numismatic Chronicle
149,
SYDENHAM.
E.A.
1933:
The Coinage
of
Caesarea in
Cappadocia
(London).
TYLECOTE,
R.F.
1986:
The Prehistory
of
Metallurgy in the
British Isles
(London, Institute of Metals).
33-58.
OXFORDJOURNALOF ARCHAEOLOGY
(c>
Basil Blackwell
Ltd.
1995
WALKER,
D.R.
1976:
The Metrology
of
the Roman Silver
Coinage. Part
I:
from Augustus to Domitian
(Oxford,
BAR Supplementary Series
5).
WALKER,
D.R.
1977:
The
Metrology
of
the Roman Silver
Coinage. Part 2:from Nerva to Commodus
(Oxford, BAR
Supplementary Series
22).
WALKER.
D.R.
1978:
The Metrology
of
the Roman Silver
Coinage. Part
3:
from Pertinax to Uranius Antoninus
(Oxford, BAR Supplementary Series
40).
WRIGHT.
R.
1991
:
Doing Multivariate Archaeology and
Prehistory
(University
of
Sydney).
77
