A Study of The Chemical Composition of Roman Silver Coinage, A.D. 196-197

A STUDY OF THE CHEMICAL COMPOSITION OF ROMAN SILVER COINAGE, A.D.
196-197
Author(s): Kevin Butcher, Matthew Ponting and Graham Chandler
Source: American Journal of Numismatics (1989-) , 1997, Vol. 9 (1997), pp. 17-36
Published by: American Numismatic Society
Stable URL: https://www.jstor.org/stable/43580403
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
American Numismatic Society is collaborating with JSTOR to digitize, preserve and extend
access to American Journal of Numismatics (1989-)
This content downloaded from

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
All use subject to https://about.jstor.org/terms
AJN Second Series 9 (1997)
© 1997 The American Numismatic Society
A STUDY OF THE CHEMICAL COMPOSITION OF
ROMAN SILVER COINAGE, A.D. 196-197

Kevin Butcher and Matthew Ponting
with a Contribution by Graham Chandler.
Introduction
The early denarius coinage of Septimius Severus was issued at a

number of different mints in the Roman empire in addition to the
mint of Rome. This phenomenon is important because it represents a
departure from the established pattern of minting seen in the second
century A.D., where denarii were generally struck only at Rome. It is
also important to note that this departure was not long term, and that
production was again centered on Rome by the beginning of the third
century. The location of the non-Rome mints has been a subject of
debate. That one of them was Alexandria in Egypt is now generally
accepted, although its output was fairly limited and was not important
in terms of the volume of coinage circulating in the empire. It operated
only for a short period, ca. A.D. 194-95. More significant are other
groups of coins, usually described as "eastern," and sometimes specific-
ally attributed to two Syrian mints, Laodicea ad Mare and Emesa,
which struck coins between A.D. 193-94 and 202. We do not propose
to discuss the attributions to mints in any detail in this article, but it
is worth pointing out that the attribution of these coins to particular
mints rests more on speculation based on historical events in the reign
of Severus than on any hard evidence, such as hoards or finds.
17

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
18 Kevin Butcher and Matthew Ponting
There is a second significant aspect to the denarius

mius Severus. In A.D. 194 he debased the denarius of Rome, esta-
blishing a new lower standard. The significance of this debasement
cannot be overstated. Its effect can be seen in hoarding patterns through-
out the empire, and it can truly be considered as a watershed between
hoarding patterns of the second century and patterns of the first half
of the third century. Hoards containing large numbers of Seveřan post-
debasement coins usually contain few pre-debasement denarii, and vice
versa. That some of the pre-debasement hoards close after the debase-
ment can be seen from some of the later strays, and we note in parti-
cular the occasional coins of the usurper Clodius Albinus, struck ca.
A.D. 196. We will not go into the possible reasons for these patterns
here, but it is important to note that the hoard evidence implies that
contemporaries saw the pre- and post-debasement coins as being differ-
ent in some way. This may be simply because they were debased, or it
may reflect the fact that the new coins had a different value compared
to, say, the gold coinage. If the post-reform coins were perceived as
debased and/or different and they were intended to form the backbone
of Rome's silver coinage from A.D. 194 onward, Severus would have
needed to put more of the new denarii into circulation quickly and
efficiently, and one interpretation of the opening up of regional mints
is that it was a response to this need.
Some of these observations will be explored in later studies of the
Seveřan silver coinage. The purpose of our investigations here is
twofold. First, to determine the silver standard used by Severus after
the debasement and to obtain a clearer picture of the metallurgy of
the Seveřan silver coinage. Second, to examine the minor and trace
element profiles of some of the different mints, not in order to try to
discover their locations, but to determine differences or similarities in
their elemental characteristics which may relate to the source of metal
and/or differences in the production technologies.
The Coins Analyzed
Four groups of coins, all of the period 196-97, were chosen fo

present study:
1) Denarii of Rome belonging to Severus's eighth impera torship (IMP
VIII);

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
Chemical Composition of Silver Coins 196-197 19
2) Denarii of an uncertain eastern mint, also of Severus' eigh

ratorship;
3) Denarii of an uncertain eastern mint, of Severus's eighth
imperatorship (IMP VIII and IMP Villi);
4) Denarii of the usurper Clodius Albinus, struck in the western
empire.
These groups were chosen with a particular relationship between the

two groups of eastern coins in mind. They are both traditionally consi-
dered Syrian, and even to be consecutive products of the same mint
(Laodicea ad Mare), although stylistically and typologically they are
quite different. The second group listed above is often called "Laodicea
old style" and the third group "Laodicea new style." A recent article
which discusses these coins restates the traditional view: "the 'old style'
series is undoubtedly Eastern, and I do not believe that it can persua-
sively be contended that the 'new style' is not a continuation from the
same mint."1 As a control it was thought useful to use two groups of
coins which are relatively plentiful and which we know are not eastern,
the products of Rome and the denarii of Clodius Albinus which were
probably struck in Gaul. These could then be compared with the
eastern coins.
The coins selected for analysis were all from the collection of the
British Museum. For assistance with this task we would like to thank
Andrew Burnett, Roger Bland, and Jonathan Williams of the Depart-

ment of Coins and Medals, who very kindly placed the relevant coin
at our disposal. The individual coins were selected to cover severa
different reverse types per group, but it should be noted that we wer
also restricted by the suitability of individual specimens for our
sampling technique.
Group 1. Septimius Severus, Rome.

1. BMC 151 ADVENTI AVG FELICIS SIMO
1 Bickford-Smith, R. A., "The Imperial Mints in the East for Sept

It Is Time to Begin a through Reconsideration," RIN 96 (1994/95),

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC

5. BMC 157 ARAB ADIAB COS II PP
6. BMC 161 FORTVNAE REDVCI
7. BMC 167 LIBERALITAS AVG II
8. BMC 169 MVNIFICENTIA AVG
9. BMC 178 VOTA PVBLICA
10. BMC 180 VOTA PVBLICA
Group 2. Septimius Severus, Laodicea old style

1. BMC 441 CERERERVGIF
2. BMC 442 FORT REDVC
3. BMC 443A FORT REDVC
4. 1979 6-14-16 FORT REDVC
5. BMC 444 FORT REDVC
6. BMC 445 LIBER AVG
7. BMC 446 LIBERTA AVG
8. 1947 10-1-10 LIBERTA AVG
9. 1946 10-4-827 MART VICT
10. BMC 449 MONET AVG
Group 3. Septimius Severus, Laodicea new style, obv. endi
1. BMC 451 HERCVLI DEFENS

2. 1979 6-14-17 PM TRP V COS II PP
3. BMC 456 PROVIDENTIA AVG
Caracalla Caesar, Laodicea new style

4. 1946 10-4-829 SECVRITAS PERPETVA
5. BMC 461 SECVRITAS PERPETVA
Septimius Severus, Laodicea new style, obv. ending IMP

6. BMC 463 PM TRP V COS II PP
7. BMC 466 PROFECTIO AVG
Group 4. Clodius Albinus Augustus, obv. legend IMP C

AVG
1. 1946 10-4-761 SPE AVG COS II

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
2. BMC 267 SPE AVG COS II

3. BMC 268 VICT AVG COS II
Obv. legend IMP CAE D CLO SEP ALB AVG

4. BMC 269 CLEMENTIA AVG COS II
5. 1946 10-4-764 FIDES LEGIONIS COS II
6. BMC 272 IOVIS VICTORIAE COS II
7. BMC 273 MARS PATER COS II
8. BMC 275 SAECVLI FEL COS II
9. 1946 10-4-760 SAECVLI FEL COS II
10. BMC 277 VICT AVG COS II
Analytical Problems
The denarius coinage of the Seveřan period has been the fo

number of analytical projects aimed at gaining a better unde
of the processes of corrosion and segregation in base-silver alloys.
Alloys of silver and copper which contain less than about 90% silver
will generally contain two phases: one rich in copper and one rich in
silver. Because of the difference in chemical reactivity (electrode
potential) the less noble (more electro-negative) phase will corrode
preferentially to the silver-rich phase. This process leads to the surface
layers of the alloy losing copper to the environment which results in an
apparent enhancement of the silver content. For coins, the problems of
natural copper-depletion are likely to have been further exacerbated
by a blanching process conducted at the mint prior to striking. This
process, involving treatment with an organic acid, would have been
necessary to remove the layer of black copper oxide which would
have inevitably formed on the freshly cast coin blank. However, it
seems unlikely that the silver enrichment effect accompanying such a
treatment2 would have gone unnoticed at the mint, especially as the
2 By the same chemical processes as undergone during corrosion, the preferential

removal of the less electro-negative copper-rich phase.

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
debased alloy would have looked very yellow unle

way. The maintaining of a currency which at lea
were made of fine silver would have been of conside
to the issuing authority, and this effect, although fortu
instance, would certainly have been exploited.3
Condamin and Picon were the first to examine the preferential
corrosion of copper in denarii of Septimius Severus4 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.
They attributed this solely to the effects of corrosion. They also
demonstrated that the metal within the coin was more truly represen-
tative of the original alloy. In particular this article demonstrated the
problems of "non-destructive" Neutron Activation Analysis (NAA) as
applied to silver-alloy coins. Used in this way, NAA 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 Seveřan denarius.5 Further-
more, it was also shown that the enriched area can often extend some
way into the coin. Despite this demonstration of serious problems with
such approaches, as well as other related work, notably by Cope,6
many have claimed to obtain accurate and meaningful results from
surface analyses.7 Carter's technique, as applied to silver-alloy coins,
was to air-abrade the area to be analyzed removing a total estimated
60 (Jim of surface metal. This was the amount of metal removed in
order to achieve a repeatable reading (three repeats starting at 20
3 Examples of blanching facilities have been found in medieval mint workshops in

Norway (McLees, L. H., "The Late Medieval Mint Workshops at the Archbishop's
Palace, Trondheim, Antiquity 68 f 19941, pp. 264-74).
4 Condamin, J., and M. Picon, "The Influence of Corrosion and Diffusion on the
Percentage of Silver in Roman Denarii," Archaeometry 1 (1964), pp. 98-105.
5 Condamin and Picon, p. 104.
Cope, L. H., "The Metallurgical Analysis of Roman Imperial Silver and Aes
Coinage," pp. 3-47, and "Surface-Silvered Ancient Coins," pp. 261-78, in Hall and
Metcalf, Eds., Methods of Chemical and Metallurgical Investigation of Ancient
Coinaae, Royal Numismatic Society Special Publication 8 (1972).
7 Condamin and Picon, above, n. 4.

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
(xm, ending at about 60 ¡Jim), the assumption being that th

sentative of the main bulk of the coin. However, it appears
misleading. There is a notable discrepancy between the s
for Seveřan denarii issued after A.D. 193 as reported by
and Picon8 (45%) and as reported by Carter.9 Carter gi
average figure of 58.4% ±0.5% (normalized) based on ten
This is a discrepancy of over 13%. Yet it is this method (or
of it) which has been repeatedly used for numismatic resear
it is this method which was used by Walker in his magnum
Metrology of the Roman Silver Coinage 11 and gave a mean s
of 57.6% for the same issues analyzed by Carter.12 This dis
clearly needs explaining and suggests that the abrasion
does not, in all cases, produce a representative analysis.
work by the authors13 has looked at Roman provincial silve
and has demonstrated that surface analyses, such as those b
invariably overestimate the silver content and also give a fal
sion of great inconsistency. Both these effects are the resul
nique which only analyzes metal from within the enriched z
Both the natural and artificial processes of enrichment h
results. However, the striking of the coin after 'blanc
compress and consolidate the silver-rich zone into a cont
compacted layer. This can clearly be seen in the structu
coins under the scanning electron microscope (SEM) (Fig. 1)
8 Condamin and Picon, above, n. 4.

Carter, G. F., "Reproducibility of X-Ray Fluorescence Analyses o
Severus Denarii," Numismatic Chronicle 19 (1978), pp. 67-73.
10 See Carter, p. 70.
11 Walker, D. R., The Metrology of the Roman Silver Coinage , p
Pertinax to Uranius Anto ni mus, Oxford, BAR Supplementary Series 40
12 Condamin and Picon analysed 54 coins of between 194 and 211 A
a figure of between 45% and 50% for 36 of the coins. Walker (197
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% (st.dev of 4 - 8) for his different issue groups within that period
13 Butcher, K. E. T. and M. J. Ponting, "Rome and the East. Production of
Roman Pronvincial Silver Coinage for Caesarea in Cappadocia under Vespasian,
A.D. 69-79," Oxford Journal of Archaelogy 14, 1 (1995), pp. 65-77.

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
Fig. 1 SEM photomicrograph of a cut section of a

alloy Roman provincial coin showing a typical tw
the interior and a mineralized copper-phase towa
gray is silver rich, light gray is copper rich, and
ized copper rich). The edge of the coin is to the
strong alignment of the phases is the result o

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
Percentage Data for Seveřan Denarii
No. iRef. 1 Au %1 Sn %| Ag%| Cu % Pb % Bi % Zn % Fe % Ni %

Septimius Severus , Laodicea 'old style,' IMP VIII
1 1 BMC 441 0.24i <0.5
10 [BMC 449 0.191 <0.5 1 48.4| 54.9

Septimius Severus, Laodicea 'new style,' IMP VIII
11 [BMC 451 0.43 i <0.

13 [BMC 456 0.301 <0.5 1 47.8 50
Caracalla Caesar, Laodicea 'new style'
14| 1946 10-4-829 0

15 1 BMC 461 0.35
Septimius Severus, Laodicea 'new style, ' IMP Villi
16 1 BMC 463 0.29i 1.05i 46.91 50.6 0.62 <0.03 0.033 0.14 0.142 98.6
171 BMC 466 0.33 0.93 47.7 49.5 0.48 <0.03 0.026 0.13 0.226 98.2
Clodius Albinus, Augustus
25 1 BMC 268 I 0.40 <0.5 84.9 13.6 0.48 <0.03 0.012 0.04 <0.01 99.4
30 [BMC 269 0.3ó| <0.5 1 73.5 1 25

Septimius Severus, Rome, obverses ending IMP VIII
31 ¡BMC 153 0.161 <0.5 1 43.51 52.7 0.60 <0.03 0.025 0.07 0.035 97.0
24 1 BMC 169

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
The zone enriched by corrosion alone will still ha

phases clearly visible, but many of the copper-r
mineralised or empty. The combination of both pr
silver rich zone of broadly two compositions exten
table distance into the coin. The consolidated blanc
the surface will be of pure silver, possibly over 90%
few microns thick. The corroded layer will be
patchy, with some areas being more corroded than
on the original structure. This will often extend to
be of a higher silver content than the original alloy
cases this corroded zone may extend throughou
making an accurate analysis impossible. Consequ
abrasion and analysis technique as developed by
by Walker will avoid the blanched layer but mis
corroded zone beneath. The results bear this out, w
lowest figures approaching the real value. Further
tency which is apparent in Walker's figures from
deviations is a result of the unpredictable depth o
and the variability of the corrosion effects.
This study reassesses the fineness of the denar
period, in the light of the above discussion, by an
metal taken as drillings from the "heart-metal" of
analytical technique has been described in detail el
Results
Silver Content and the Problem of Standards
The silver contents for the Seveřan issues agree very well wit
findings of Condamin and Picon, being around 46%, and confir
unreliability of Walker's figures for the reasons discussed abov
silver content of 46% is also rather more consistent than Walker's 55-
60% figure, having a relative standard deviation (RSD) of 2.8
compared to 10.5 for Walker's figures. The contemporary denarii of
Clodius Albinus also have a lower mean silver content of 78.4%
14 See Butcher and Ponting, pp. 67-69.

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
compared to Walker's 89.2%. However, this does not affec

conclusion that Albinus's denarii were struck on the earlier Flavian
standard. Our reanalysis of Flavian denarii has shown Walker's figures
to be equally inaccurate with 78.7% being the newly estimated mean
value.15
Silver metal as perceived in antiquity was very different from
elemental silver as measured by modern analytical techniques. The
processes of smelting and refining silver from its various ores could
not produce chemically pure silver. Traces of other elements present
in the original ores were inevitably carried through the processes and
remain to be found in the coins themselves. Silver in antiquity was
commonly produced from argentiferous lead ores or from other ores
using processes which required the addition of lead.16 In either case
argentiferous lead metal was produced, where the silver occurs as a
minor component in the lead together with minute traces of gold and
bismuth, etc. coming from the original ore. In order to extract the
silver from the lead a second process called cupellation was used. This
involved heating the metal to about 1000° C in a strong air flow which
oxidized the lead and most other impurities. Only the elements not
readily oxidized would remain in their metallic state. These consisted
primarily of the silver and any gold in the original ore which also
would have been carried through into the silver. Certain other
elements would also have been difficult to remove by cupellation,
notably bismuth. Consequently any silver produced by this process
will contain traces of gold similar in concentration to the original ore
and possibly traces of bismuth dependent on ore type used. Traces of
lead will also invariably remain as complete cupellation was rarely
achieved, especially in large scale operations as were the case in the
Roman period. The silver which would have been used to make the
coinage alloy would therefore contain these trace impurities. Roman
silver is generally considered to contain between 0.05% and 2.5%
lead, 0.1% to 0.5% gold and, generally, 0.01% up to 1% bismuth.17
15 See No. 13. (1995) pp. 67-69. The fact that in this case Walker's conclusions
are valid should not lead the reader into thinking that in general his conclusions
about comparative silver standards are valid.
16 Craddock, P. T., Early Metal Mining and Production (Edinburgh, 1995), p. 213.
17 See Craddock, p. 213.

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
This means that up to 3% of the metal which was added to the

melting pot as "silver" was not silver. It therefore seems sensible to
regard the fineness of an ancient coinage as being the combined
silver, gold, lead, and bismuth contents. However, by a small scale
cupellation or "fire assay" virtually all the lead can be removed, and
this may have been the technique employed to establish the purity of
silver in antiquity, as was certainly the case by the twelfth century.18
It is therefore likely that the silver refined on large scale cupellation
hearths (down to around 98% fine silver)19 would have been known
to be less than pure by the fire assaying, if this was used as a quality
control. The question then is whether the standard on which the
coinage was struck took this fact into consideration. Certainly this
was the case in medieval England, where the sterling standard was
calculated as 92.5% (silver plus gold) and excluding any lead present
from large scale cupellation.20 However, if we calculate the fineness of
the Seveřan denarii as being the sum of the silver, gold, and lead we
arrive at a figure close to 50%. The use of a 50% standard would
seem very logical compared to a 48.5% or 49% standard and would
suggest that the presence of 1% - 2% lead, even if known about, was
not deemed sufficient to detract from the required purity of the stan-
dard. Given the high level of base metal in the Seveřan denarii it is, of
course, possible for some of the lead to come from the copper.
18 Pliny does not refer to fire assay specifically, but small scale cupellation of the
type is discussed by Percy (1870) in his book The Metallurgy of Lead (pp. 209-11)
was able to produce feinsilber >99% pure and therefore act as a method of checking
larger scale operations. This process is certainly of some antiquity, being described
in some detail by both Agricola (de Re Metallica Libri xii, Basileae, 1561, pp. 385-
492) and Erker (see n. 19 below) as well as having some archaeological basis, for
example the small scale cupellation hearths found at Silchester and Hengistbury
Head (above, n. 12, see Craddock p. 229). Whether the small scale cupellation
process was used habitually as a final stage to produce pure silver or solely as an
occasional method to check the purity of the larger scale cupellation product is
unimportant. What is important is knowing that this stage is necessary. There is
no definite evidence that fire assaying was used during the Roman period, but the
technology was certainly available.
19 Siseo, A. G., and C. S. Smith, Translation of Lazarus Erker's 1580 Treatise on
Ores and Assaying (1951), p. 64.
20 Ponting, M. J. (forthcoming), "Analysis of the Coinage of Henry 1st," appendix
in M. Archibald and W. Conte, British Museum Occasional Publication.

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
However, this is likely to be very small given that Roman

rally seem to have used freshly smelted copper and not recycled
scrap.21 This view is also supported by the low levels of other trace
elements which would otherwise indicate the use of recycled base-
metal. The one exception to this is with the later Laodicea New style
coins of Severus, which have a significant trace of tin present, and
may suggest the use of recycled metal containing low levels of tin.
The denarii of Septimius Severus therefore appear to have been
struck on a standard of approximately 50% silver with the remainder
generally being virtually pure copper. The denarii of Clodius Albinus,
on the other hand, were struck of purer metal, probably on a standard
of 80% silver, the same as that of the Flavian issues. Both these
figures represent easily calculable amounts by the methods and termi-
nology of the time. The 50% standard would be calculated as the
silver being alloyed with one part in two of copper, and the 80% as
being alloyed with one-fifth part of copper. The silver contents of the
issues investigated are shown as box and whisker plots in Fig. 2, which
also show the spread of values.
loo-
90 > _____
-m 80 «
? r_ JL_
i
+
§ 60-
+
OC „
a 50 „ 1
93 P-
^
r 40.
>
E 30 J
N- 10 4 3 1» 10
Laodicea old Vili
Type catego
Fig. 2. Fi
21 Ponting, M
Roman Coppe
Britain," Ph.

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
The larger spread apparent in the Laodicean old st

one particularly base coin, but the consistency of t
coins is very clear. The coins of Clodius Albinus, on
are somewhat less consistent and this may reflect,
the minting technology (blanching, etc.) or in the q
silver from which the alloy was made.
Minor and Trace Constituents
The only element apart from silver and copper to occur at levels of
1% and above is tin. This is unusual in Roman silver coinages and is
of interest as it only occurs with any consistency in one particular
group, the Laodicea new style coins of IMP Villi and the two coins of
Caracalla Caesar, ascribed in BMC to the period of Severus's Laodicea
new style IMP VIII coinage. The most probably explanation for this is
that recycled low tin bronze was used as the alloying component.
Gold, lead, and bismuth are potentially the most useful elements in
distinguishing between the products of different mints for the reasons
discussed above. A plot showing the gold and lead contents expressed
as a ratio with the silver (to allow comparison across the two different
silver contents) clearly shows the Rome issues of Severus as being
quite distinct from the issues of Lyon and the east (Fig. 3). The
Rome issues have a significantly higher level of lead, calculable as
1.1% in fine silver, whereas the Lyon issues of Clodius Albinus and
Severus' Laodicean coins have only about 0.5% lead in fine silver.
This probably represents a difference in the cupellation processes and
can be used to suggest two different models.
1. Silver produced in the east, and in the western provinces controlled

by Clodius Albinus, was the product of smaller scale refining
processes more likely to remove a greater proportion of the
remaining lead. Or
2. These mints relied more heavily on recycled silver which had been
repeatedly recycled over many (hundreds of) years and would there-

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
Fig. 3. Silver: Gold and Silver:Lead Ratio Plot
fore have been the product of smaller scale refining pro

is assuming straight re-use of metal, without any refi
would have involved the addition of lead. The second model is,
perhaps, the most tenable, especially as the gold contents are gene-
rally quite homogeneous. This would suggest a supply pool of silver
which had been homogenized by repeated remelting and recycling.
The bismuth concentrations, expressed in the same way and plotted

against the silver:gold ratios (Fig. 4) clarify the picture. The separation
of the Rome and Lyon issues is confirmed. However, a separation is
also suggested between the Laodicean old and new styles, with the
new style coins generally overlapping the issues of Lyon. Both Rome
and Laodicea old style form quite distinct clusters (with one or two
outliers) due to their higher bismuth content, whereas the other issues
are generally spread out.
22 Analysis of Roman and Sassanian silver plate has shown a similar difference,
with Roman plate generally containing l%-2% lead and Sassanian silver plate gene-
rally containing 0.5% lead (Hughes, M. J. and J. A. Hall "X-Ray Fluorescence
Analysis of Late Roman and Sassanian Silver Plate," Journal of Archaeological
Science 6 [1979], pp. 321-44).

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
Fig. 4 SilveriGold and Silver:Bismuth Ratio
The suggestion that the Laodicea old and new st

tionally different can be taken further using a sta
nique called discriminant analysis. This statistical
validity of a grouping solely on the mathematica
compositional variables and gives a probability value that each case
(coin)- is correctly grouped. Using only the trace elements present, a
discriminant analysis correctly attributed 97.3% (39 out of 40) of the
coins to their issue groups. This is a very good result and indicates
that there are significant chemical differences between all the issue
groups. A plot of the discriminant functions illustrates this (Fig. 5),
showing that the most similar groups are Severus Rome issues and
Clodius Albinus's Lyon issues. However, these are groups that we
already know clearly differ from each other in their trace element
concentrations and serve to illustrate the even greater difference
between these and the Laodicean issues. The old and new style coins
are well separated and the very different coins of the SECVRITAS
PERPETVA type and IMP Villi are even more distinct. A closer study of
the discriminant statistics shows that, apart from lead, gold and
bismuth, iron and nickel concentrations play a significant role in the
issue groupings.

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
Fig. 5. Plot of First Two Discriminant Functions
This is interesting as these two elements are strongly ass

the copper in the alloy, whereas the others are associated
silver. Nickel tends to relate directly to the source of the
with some copper ores containing more nickel than others
iron relates directly to the smelting and refining technolo
speaking, the more highly refined the copper, the lower the iron
content23 or the more frequently remelted (some iron being lost as
oxide in every melt). The highest iron contents are found in the new
style issues of Laodicea (mean of 0.1%) (Fig. 6) with the Rome (mean
of 0.05%) and old style issues overlapping. However, the old style
coins generally seem to have the lowest iron contents (mean of 0.03%
with several coins below detection limits). It generally appears that the
Roman mints used freshly smelted copper which was refined only
enough to render it workable, consequently iron content of 0.1% to
0.5% are common in copper-based coins.24 The higher iron content
found in the Laodicean and Rome issues would be broadly consistent
23 See n. 13, pp. 73-74, and n. 16, pp. 137-41 for a more detailed discussion.
24 See above, n. 21.

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
•161
•14, i i
•12, - ■ - ,
.10.
.08 1
.06-
.04 « 1
Í 02. I - L_
ê 0.00,
i /=
old style IMP VIII new styme IMPVIII new style IMPVIIII Rome, IMP VIII
Issue
Fig. 6. Iron Content of Seveřan Denarii
with this level of refining. The lower levels may suggest the use of
recycled metal or merely differences in refining standards. The nickel
levels are also higher in the new style Laodicean issues (Fig. 7) with a
mean of 0.2% against a mean of 0.09% for the old style issues (and an
even lower figure of 0.02% for the Rome issues).
•41
*12
.3.
.2« ____
^ .1. ____
______
"n= 'to ? ! 15
old style IMP VIII new style IMPVIII new style IMPVIIII Rome, IMP VIII
Issue
Fig. 7. Nickel Content of Seveřan Denarii

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
Conclusion
The standard of the post-reform Seveřan denarius is now establish

as having been 50% cupelled silver and the standard of the coinage
Clodius Albinus issued at Lyon was 80%. This differs significantly
from the analyses published by Walker, but does not alter the validity
of his conclusion that Clodius Albinus was issuing denarii on the
Flavian standard. The coinage of Clodius Albinus Augustus is an
important phenomenon which deserves further examination: the
hoarding pattern observed in the introduction to this article may be
explained by the fact that the silver standard of these coins was
perceived by contemporaries as being equal to the pre-reform denarii
of the late first and second centuries A.D., and that these coins were
different from the post-reform denarii. The post-reform Seveřan stan-
dard of 50% is also demonstrably more consistent than Walker's
higher standard, as would be expected from the metallurgical study
of the minting processes and analytical techniques employed. This
50% silver standard is encountered in other Roman coinages of the
first and second centuries A.D., which we have examined elsewhere.25
These coinages, however, are all Roman provincial coinages, issued in
the eastern provinces. Nevertheless, one of the authors has previously
suggested a link between this provincial standard and the adoption of
the 50% standard for the denarius under Septimius Severus.26 We
hope to be able to examine this possible link at a later date.
The denarii of Clodius Albinus not only differ from those of Severus
in their silver content, but also in their trace element concentrations.
In particular the bismuth concentrations are generally lower,
suggesting the use of different silver sources. The issues of Rome and
Laodicea also differ on the very basic level of their silverrgold and
silver:lead ratios. However, the Laodicean new style is also compositio-
nally different from the old style on two fundamental levels.
1. The bismuth contents of the old style issues are generally higher,
and
25 See Butcher and Ponting, above n. 13, pp. 68.

26 Butcher, K. E. T., "Rhodian Drachms at Caesarea in Cappadocia," Numismatic
Chronicle 1992, pp. 41-48.

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC
2. The iron and nickel levels are significantly hig

coins than in any of the other groups, with th
coins generally having the lowest iron concentra
This gives two independent arguments for the

new styles being the products of two different mi
levels indicate the use of silver from different sou
and nickel levels indicate not only the use of cop
sources, but also significant differences in the lev
during copper production. This suggests that the
style coins was generally less well refined and
freshly smelted. From these results we cannot co
Laodicean old style and Laodicean new style mint
sufficient evidence to challenge the traditional su
the same mint. The most notable compositional group is that
comprising the two examples of the Laodicean new style issues of IMP
Villi and the two Caracalla Caesar coins. The compositional similarities
suggest that the SECVRITAS PERPETVA coins of Caracalla should be
placed with Severus's coins of IMP Villi (or perhaps with later IMP X
coins of A.D. 197 which were not analyzed), although the sample is
very small and further analyses might uncover IMP VIII coins with
similar characteristics. Nevertheless, these coins must be seen as a
distinct compositional group and point to a marked change in mint
practice.
Acknowledgments
In addition to the staff of the British Museum mentioned above, the

authors would like to thank Dr. D. Griffiths and Dr. J. Merkel of the
Institute of Archaeology, University College London, for allowing

access to analytical equipment. This study is part of a larger series of
analyses of Roman silver coinage, and the authors gratefully acknow-
ledge the support of the British Academy Fund for Applied Science in
Archaeology and the Society of Antiquaries of London for respective
grants to meet the costs of analysis and study. In addition we should
like to thank colleagues overseas, particularly Arthur Houghton, Rick
Witschoncke, and David MacDonald, for their enthusiastic support of
the project.

132.174.254.12 on Sat, 11 Jun 2022 23:58:00 UTC

A Study of The Chemical Composition of Roman Silver Coinage, A.D. 196-197

Uploaded by

Copyright:

Available Formats

A Study of The Chemical Composition of Roman Silver Coinage, A.D. 196-197

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Study of The Chemical Composition of Roman Silver Coinage, A.D. 196-197

Uploaded by

Copyright:

Available Formats

A STUDY OF THE CHEMICAL COMPOSITION OF ROMAN SILVER COINAGE, A.D.

Stable URL: https://www.jstor.org/stable/43580403

This content downloaded from