An adapted method for researching ancient Egyptian mirrors

Abstract

Some metallurgical analyses have been conducted on Ancient Egyptian mirrors; however, both compositional and microstructural data are necessary in order to fully reconstruct the manufacturing sequences of these artefacts. Traditional sampling and analytical methods for researching metal artefacts have their limitations for investigating mirrors in particular; for example, a mounted 'V' cross-section is often not a viable sampling option as it is too visually destructive to a complete disk, and surface analysis only provides limited compositional data. These observations resulted in the adaptation of deep-filed edge abrasion sampling, based on coin studies, with SEM-EDX analyses for Egyptian mirrors. This paper will establish how the methodology gathers reliable compositional and microstructural data while remaining visually discreet. This study demonstrates the methodology on four mirrors varying in condition, shape, and size in comparison to the traditional mounted 'V' cross-section taken in the 1990s from the same specimens.

Full text

An adapted method for researching ancient Egyptian mirrors
Elizabeth Thomas
*
, Peter Gethin
University of Liverpool, Department of Archaeology, Classics and Egyptology, 12-14 Abercromby Square, L69 7WZ, United Kingdom
ARTICLE INFO
Keywords:
Methodological
SEM-EDX
Metallurgy
Artefacts
ABSTRACT
Some metallurgical analyses have been conducted on Ancient Egyptian mirrors; however, both compositional
and microstructural data are necessary in order to fully reconstruct the manufacturing sequences of these ar-
tefacts. Traditional sampling and analytical methods for researching metal artefacts have their limitations for
investigating mirrors in particular; for example, a mounted ‘V’cross-section is often not a viable sampling option
as it is too visually destructive to a complete disk, and surface analysis only provides limited compositional data.
These observations resulted in the adaptation of deep-led edge abrasion sampling, based on coin studies, with
SEM-EDX analyses for Egyptian mirrors. This paper will establish how the methodology gathers reliable
compositional and microstructural data while remaining visually discreet. This study demonstrates the meth-
odology on four mirrors varying in condition, shape, and size in comparison to the traditional mounted ‘V’cross-
section taken in the 1990s from the same specimens.
1. Introduction
There has been signicant attention given to Ancient Egyptian mir-
rors within the literature. The work of scholars such as but not limited to
Lilyquist (1979), Derriks (2001), Casta˜
neda Reyes (2010) and most
recently Odler (2023) go into great detail regarding their archaeological
context, depictions and religious connections to the sun God Ra or
Goddess Hathor, and their ties to the elite women of Egyptian society,
beauty and fertility etc. However, the metal disk of the mirror itself has
received minimal consideration, with comments generally being limited
to ‘made of bronze’or ‘polished to almost gold’(Szpakowska, 2008).
Such statements can be misleading as archaeometallurgical analyses
have demonstrated that Egyptian metal-workers used both arsenical
coppers and tin bronze (Ben-Yosef, 2018) and that these can have a
silvery appearance (Meeks, 1993), for example.
Metallurgical studies that have included Egyptian mirrors often only
focus on the chemical composition of the metal, which, while informa-
tive does not fully answer questions related to their manufacture as only
microstructural investigation can do this. The limited data is partly due
to the research questions of previous investigations which inuenced
their sampling and analytical techniques. In order to gain microstruc-
tural information, traditionally, a ‘V’cross-section (VS) would be
removed from an inconspicuous location on the specimen and mounted
in resin. However, this type of sampling is often deemed too destructive
by several stakeholders for intact whole artefacts, and therefore unlikely
to be permitted. Archaeological scientists also do not always deem this
practice appropriate, such as when there is no real inconspicuous area to
sample from. Other options such as the use of surface analytical tech-
niques do not provide microstructural information or representative
compositional data, particularly if the specimen is severely corroded
(Scott, 1991).
Investigations into Egyptian mirror manufacture have not
completely addressed research questions regarding microstructure,
particularly related to the production of surface enrichment (M¨
odlinger
and Sabatini, 2016: 72). This observation by the authors resulted in the
adaptation and development of a minimally destructive sampling and
analytical methodology, which offers a viable alternative to a mounted
‘V’cross-section whilst still gathering reliable compositional and
microstructural data. The sampling technique deep-led edge abrasion
(Cope, 1974: 70), which produces a tangential taper section (TTS)
(Anheuser and France, 2002), is based on work conducted in numis-
matics (King and Northover, 1993). Although coins and mirrors differ in
size, they are similar in shape. Therefore, it was thought that this
method of sampling could be applied to a mirror disk. This investigation
provides a comparative study of the adapted methodology with a
traditional metallurgical sampling technique, a mounted ‘V’cross-
section.
* Corresponding author.
E-mail address: hsethom5@liverpool.ac.uk (E. Thomas).
Contents lists available at ScienceDirect
Journal of Archaeological Science: Reports
journal homepage: www.elsevier.com/locate/jasrep
https://doi.org/10.1016/j.jasrep.2024.104743
Received 21 March 2024; Received in revised form 26 July 2024; Accepted 26 August 2024
Journal of Archaeological Science: Reports 59 (2024) 104743
Available online 12 September 2024
2352-409X/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ).

2. Background
One of the largest catalogues of Ancient Egyptian mirrors was pro-
duced by Lilyquist (1979), where some compositional data is present,
but there is no accompanying in-depth metallurgical discussion since
that was not the objective of the work. While a substantial collection of
40 mirrors from the Louvre Museum were analysed using Atomic
Emission Spectroscopy by Michel (1972), other published metallurgical
investigations have included a limited number of mirrors. When they
were included they generally formed a small part of larger assemblages,
for example out of 43 artefacts, a single mirror from the Middle Kingdom
was analysed in Gilmore’s (1986) research into copper objects from
Kahun, where a drilled sample revealed the mirror to be made of
arsenical copper.
Much of the metallurgical research so far into Egyptian mirrors has
primarily focused on composition for example, Masson-Berghoff et al.
(2018), where drilled samples of core material were analysed for their
study consisting of only two mirrors. Also, Rademakers, et al. (2021)
analysed 4 mirrors from the Middle Kingdom, showing a mixture of alloy
types were being used at this time with both arsenical copper and
arsenical tin bronze being identied. While this data is still useful, that
alone is unable to tell us about production methods, only the materials
involved. To answer questions about production, microstructural anal-
ysis is needed.
A further six mirrors dated between the Old Kingdom and Middle
Kingdom were included in Eaton and McKerrell (1976), where one side
of the disks were covered in a silvery layer which had a high arsenic
content. They utilised X-Ray Fluorescence (XRF) for surface analyses
only but there was no discussion about the removal of corrosion prior to
analyses. The concentration of an alloying metal can differ in the
corrosion across the surface of an object, as exemplied in Odler et al.
(2018: 423-424), where the amount of arsenic present on the surface of a
mirror ranged between 3.4 and 4.8 wt% but doubled when a different
spot was analysed. This shows how surface analysis is not an accurate or
reliable method of measuring the composition; sufciently exposed bulk
metal from underneath the corrosion layers is a necessity. On the other
hand, Rademakers et al. (2018) removed corrosion using a steel brush to
ensure they could obtain a metallic sample, but the process could also
have removed evidence of surface enrichment as this layer can be only a
few microns thick (Meeks, 1993: 263). As such, research questions
related to the mode of manufacture in particular require a cross-section
of the material, to ensure not only accurate compositional data is
gathered but also that potential surface enrichment can be identied.
Furthermore, the above also demonstrates how the location from
which a sample could be taken may not always be representative of the
object as a whole. For mirrors, the most inconspicuous location to take a
taper section is the bottom of the tang. However, the collection of
Egyptian mirrors at the Garstang Museum, University of Liverpool was
sampled in the 1990s (Simpson, 1993) via a ‘V’cross-section in this
location. When examining the corresponding mirror disks to the samples
there was often a macroscopic difference in thickness between the tang
and the disk. This would mean microstructural data gathered from the
tang would not be representative of the disk so any conclusions would
not be applicable to the mirrors as a whole. From this, it is clear there is a
need to sample from the disk directly but this must be done in the most
discreet way possible while still gathering all relevant data.
The microstructure of Egyptian mirrors has been touched upon by
Odler and Kmoˇ
sek (2020) and Garenne-Marot (1984). However, only
three of the fourteen mirrors analysed in Odler and Kmoˇ
sek (2020) had
their microstructure observed, this was due to limited permissions from
the museum of which artefacts they were allowed to sample in such a
way that would allow for metallography (Odler, Personal comms.).
Garenne-Marot (1984) details how the ancient metalworkers were able
to identify the natural segregating properties of arsenical coppers and
manipulated it in order to produce silver surfaces on the mirrors.
However, no micrographs are offered in support of this statement. These
studies provide a starting point for the microstructure of Egyptian mir-
rors but do not comprehensively investigate the mirrors’mode of
manufacture.
The limited microstructural data and subsequent lack of information
about the potential of surface enrichment is exemplied by M¨
odlinger
and Sabatini (2016: Tab.3, 72). Here it is suggested that the process of
Fig. 1. Images of the mirrors. Top Left) E946, Top Right) E1452, Bottom Left) E1519, Bottom Right) E1541.
E. Thomas and P. Gethin Journal of Archaeological Science: Reports 59 (2024) 104743
2

cementation was used to produce the arsenic-rich surfaces on Egyptian
mirrors identied by papers like Eaton and McKerrell (1976), but it was
stated that there is no microstructural data to conrm such a theory.
Micrographs are crucial as they allow the identication of fabrication
methods and enable the correct interpretations of how surface enrich-
ment was produced (Meeks, 1993: 251). Therefore, it is clear there is a
need for a sampling and analytical method that permits the compre-
hensive and consistent collection of both compositional and micro-
structural data that will be permitted by museums by minimising the
observable damage to an artefact.
3. Artefact description
The assemblage consists of four mirror disks (Fig. 1) each in various
states of preservation; they are all from the Garstang Museum of
Archaeology, University of Liverpool. This selection demonstrates how
the methodology can be applied to mirrors with a range of sizes, shapes
and conditions; for example, mirror E1541 has textiles adhering to the
surface and E1452 is concave in shape. The mirrors in this assemblage
do not have any provenance information. However, they were selected
as they had been previously sampled in the 1990s (Simpson, 1993)
allowing for a direct comparison between the two different sampling
methods. The dimensions of the mirrors and details of any known con-
servation work carried out can be seen in Table 1.
4. Methodology
The following methodology can be adapted depending on the size,
shape, and condition of the individual artefact; this is a general outline
of the procedure. Also, this methodology focuses on sampling for the
retrieval of data regarding bulk composition and microstructure, as such
it would be inappropriate for trace element or lead isotope in-
vestigations, needed for metal provenance.
A scanning electron microscope with an energy-dispersive X-ray
spectrometer (SEM-EDX) was used as the primary analytical technique
for this work. The University of Liverpool Archaeology Department
operates a JEOL JSM-IT300 SEM with a Bruker XFlash6-30 EDX system.
This piece of analytical equipment was selected for a wide range of
reasons. The main advantage of using this specic SEM is that it has a
large chamber (up to 200 W ×200D mm), big enough to accommodate
whole specimens without the need for sub-sampling. In conjunction with
additional stage tilting capabilities of up to 90 degrees, the instrument
facilitated the adaptation of the minimally destructive sampling and
analytical method. Furthermore, SEM-EDX is non-destructive in nature
while producing high-resolution and high-quality greyscale images as
well as compositional information. Precise qualitative and quantitative
data for both major and minor elements in the bulk metal can be
routinely gathered (Goldstein et al., 2017). As the measurement of trace
elements was not a concern in this investigation, the sensitivity of this
analytical technique is sufcient for the level of compositional infor-
mation required. Conversely, if trace elements were of interest in a
different investigation then another sampling method such as a drill
sample may be more appropriate. The high magnication capability
coupled with a large depth of eld (Bayley et al., 2008) allows for the
identication of phases within the material such as inclusions, while the
system can also reveal the distribution of chemical elements on the
surface of the sample.
As for specimen selection; the size of the SEM chamber limits the
potential pool of artefacts to mirrors smaller than 200 ×200 mm in size
as that is the maximum capacity. This restriction would vary depending
on the SEM model and therefore the chamber size used. The state of
conservation of the mirror has to be taken into consideration as to
whether it can withstand the preparation process and placement on the
SEM stage. An aspect of this is the level of corrosion as when heavily
corroded the artefact can become fragile. There is also the question of
the extent of original bulk metal left within the core of the object and the
quantity of material needing to be removed in order to reach it, whether
it would be deemed too destructive to abrade to the required depth. On
the other hand, some museum curators in the past have taken measures
to “restore”mirrors to their original state by stripping away the corro-
sion layer with acid, this treatment results in the core metal already
being exposed so a smaller quantity of material would have to be
removed. However, the outer layer of metal on the mirror in conjunction
with the corrosion could possibly show evidence of surface enrichment;
this information may have been removed with this conservation method
which needs to be taken into consideration when discussing results. It is
typically determined by the curator or conservator of the museum as to
whether a mirror is stable enough to be sampled and analysed.
Before sampling is conducted the mirrors are photographed,
measured and weighed. For mirrors with textile remains present on the
surface, the textiles are imaged using the non-destructive Keyence VHX-
7000 4K digital microscope. The colour images of the textiles were taken
at 100x magnication in an area of 40 ×40 mm, an example can be seen
in Fig. 2. This is not an essential step as access to a Keyence microscope is
not universal however, having detailed documentation of the textiles is
useful as a precautionary measure in case of any damage occurring
during sampling. Although to date, no damage has occurred due to the
measures in place described below. Furthermore, the images could later
be analysed by a textiles expert to extrapolate further data such as warp
and weft or thread count (Gleba and Price, 2012).
The minimally destructive sampling technique used in this method-
ology is adapted from numismatics. It is known as deep-led edge
Table 1
Dimensions of the mirrors and condition information.
Accession
Number
Dimensions
(mm)
Thickness
(mm)
Weight
(g)
Condition
E946 133 x 130 2.4 188 Stripped −Cleaned and coated with Ercalene, 1969.
E1452 120 ×120 1 76 Cleaned by Liverpool World Museum, 1961.
E1519 96 ×107 1.6 95 Stripped −Cleaned by zinc and caustic soda, Sept 1957 –Over-dried
E1541 130 ×120 1.5 224 Textiles Present
Fig. 2. Keyence image of the textiles on the surface of mirror E1541.
E. Thomas and P. Gethin Journal of Archaeological Science: Reports 59 (2024) 104743
3

abrasion (Cope, 1974) which removes the surface corrosion from the
very edge of the disk, exposing the original bulk metal underneath,
through grinding and polishing (Cope et al., 1997) (Fig. 3). This type of
sampling provides a cross-section of the mirror known as a tangential
taper section (Anheuser and France, 2002).
Deciding where to sample is made on an individual basis in
conjunction with conservators from the respective museums. This pro-
cedure ensures the area selected will be the least damaging to the overall
appearance of the mirror, and as such the areas selected on each mirror
were variable (Fig. 4). Typically, the most curved point was selected as
this minimised the amount of material removed from the object.
Generally, it is advised to avoid sampling an area around the tang, as
when positioned on the SEM stage the sampled area needs to be the
highest point. If the chosen area is near the tang this would not be
achievable since the tang would be higher than the abraded area. This
also runs the risk of the tang colliding with the pole piece within the SEM
chamber causing damage to both the mirror and the equipment. Addi-
tionally, as previously mentioned, there is also the possibility that the
tang was treated differently during manufacture so sampling from this
area would not provide representative data for the disk.
The size of the exposed area varies from mirror to mirror (Fig. 5), this
is dependent upon the steepness of the curve and the size of the disk as
well as the extent of corrosion. The larger the exposed area becomes the
greater the destructiveness; however, enough material needs to be
removed in order to reveal a sufcient area of the bulk metal underneath
the corrosion. For example, a disk that has been previously stripped of its
corrosion is more likely to have a smaller sample size as it is possible to
expose a sufcient amount of the bulk metal more quickly. However, a
disk that is heavily corroded may require more material to be removed
in order to reach the intact metal. As such, an ideal sample size cannot be
stated in terms of obtaining the required data as it is solely dependent
upon how much metal remains.
A sampled area needs to be consistent in size and uniformly at for
the best results from the SEM-EDX, whilst also minimising the amount of
material removed. In numismatic, the coins are mounted on edge in
purpose-built holders leaving the desired area exposed for sampling
(Cope et al., 1997: 69). As it would have been difcult to produce a
holder that would be big enough and adaptable to hold the variety of
mirror sizes, instead a clamp was made to secure the mirror disk in place
during the grinding and polishing process (Fig. 6). Manufactured from
nylon, the clamp is both non-contaminating and durable so is resistant to
the abrasive nature of the sampling, while also being adequately soft so
as to minimise pressure damage to the artefact. Some mirrors may be
structurally weak due to the extent of corrosion which means they are
more likely to be damaged during sampling; the clamp provides some
structural support by securing the mirror in place. In order to minimise
the potential damage to the mirror from contact with the clamp, open-
cell polymer foam was placed between either side of the disk and the
clamp. If textiles are present, additional layers, rst of tissue paper and
then Paralm are wrapped around the disk. This is so that any move-
ment or contact with the clamp would not damage the textiles. Also, if
possible, clamping directly onto the textiles should be avoided. When
Fig. 4. Image showing the location of the samples taken on each mirror. Red =‘V’cross-section. Green =Tangential taper section. Top Left: E946, Top Right: E1452,
Bottom Left: E1519 and Bottom Right: E1541.
Fig. 3. Visual comparison of a tangential taper section vs a ‘V’cross-section.
E. Thomas and P. Gethin Journal of Archaeological Science: Reports 59 (2024) 104743
4

Fig. 5. The extent of exposure of the mirrors. 1st) E946 at 10.92 mm, 2nd) E1452 at 11.45 mm, 3rd) E1519 at 6.81 mm, 4th) E1541 at 12.33 mm.
Fig. 6. Diagram of the position of the mirror in the clamp.
Fig. 7. Diagram of how a mirror in the clamp is sampled.
E. Thomas and P. Gethin Journal of Archaeological Science: Reports 59 (2024) 104743
5

the mirror is placed in the clamp, only the edge of the disk that will be
prepared is exposed. The bolts are tightened enough so that the mirror
does not move –the pressure exerted by the clamp on the mirror varies
between mirrors based on their shape, thickness and weight. This set-up
ensures no material other than that exposed is removed thereby
restricting the size and depth of the prepared area, which in turn min-
imises the impact of the process. The mirror is not removed from the
clamp until the sampling process is complete.
Initially, to remove the corrosion products, the edge is ground using
1200 grit corundum abrasive paper on the slowest rotational speed
setting (Fig. 7). The nest grade paper and slowest rotating speed gives
greater control over the depth of material removed in addition to the
placement in the clamp. When a sufcient area of the bulk metal un-
derneath the corrosion is revealed the sample is cleaned using ultra-pure
water to remove contaminants. Next, the sample is polished using 6-µm
diamond paste and as soon as the polish at this grade is sufcient it is
cleaned and the polish is moved down a grade; to 3 µm and eventually 1
µm. Both grinding and polishing are conducted using water which acts
as a lubricant and coolant. When complete, the mirror is removed from
the clamp, cleaned with ultra-pure water to remove any remaining
polishing paste, then ethanol and dried. For the majority of mirrors to
date, the corrosion has not been friable but a stable patina. As such, the
introduction of water does not weaken the corrosion layers however,
this may be different if there is the presence of bronze disease.
Furthermore, the use of ethanol and then the mirror being warm air-
dried drives off any liquid present on the surface leaving the corrosion
how it was before sampling. Additionally, when the mirror is placed
under vacuum in the SEM this removes any remaining moisture on the
mirror leaving the sample completely dry and stable.
Once sampled, the mirror is then placed in the SEM for the collection
of micrographs and compositional data alongside elemental mapping.
The position of the mirror within the SEM chamber is crucial as the
exposed edge has to be at and directly underneath the detector in order
to be in view; to guarantee this, the mirror is placed parallel to the stage
with the exposed edge protruding over the top of the stage (Fig. 8). The
Fig. 8. Diagram showing how the mirror is positioned within the SEM chamber
for analysis.
Fig. 9. Stage positioning of the mirrors. Top) E1452, Middle Left) E1541, Middle Right) E1519, Bottom) E946.
E. Thomas and P. Gethin Journal of Archaeological Science: Reports 59 (2024) 104743
6

stage is then tilted close to 90 degrees bringing the sample into view.
Ensuring there is no collision between the protruding edge and the de-
tector, the stage is lowered to the bottom of the chamber, known as the Z
axis, which for the instrument used in this investigation is 80 mm. Also,
the Y axis that controls the height of the stage when tilted must be as low
as possible (50 mm). The stage is then tilted in increments until the edge
appears to be directly beneath the detector (close to 90 degrees). The Y
axis is then raised to a working distance of around 10 mm which is the
optimal distance for gathering compositional data and clear images for
this instrument. This number may differ on other systems. This stage
conguration shows a repeated angle of between 87 and 90 degrees is
needed for the sample to be in view.
To prevent damage from contact with the metal stage, the mirror is
placed on top of nylon M4 extension nuts (Fig. 9). The nuts also support
the weight of the mirror. For the specimen to be in view in the SEM, the
stage must be tilted so nylon washers and screws are then put in place to
secure the mirror’s position. As a result, when the stage is tilted the
mirror does not move. Again, nylon was the material of choice as when
under vacuum within the SEM chamber it does not change i.e. expand or
contract, ensuring the mirror does not move or incur damage. This
method of stage positioning is quick to assemble and easily repeatable,
meaning that as long as the location of the bolts are recorded, the same
mirror could be placed back within the SEM at a later date and the area
in view would be in the same position. Also, this method can accom-
modate a variety of sizes and shapes as seen with the concave mirror
E1452 through the use of varying lengths of nylon M4 nuts (Fig. 8). It is
also possible to add or remove points of contact or support on an indi-
vidual basis so if one mirror is heavier then more bolts can be used to
secure it in place. As the mirror does not have direct contact with the
stage and the nylon bolts are not conductive, there is the issue of
charging during analysis. To overcome this, a piece of aluminium tape is
placed between the disk and stage providing a connection for the energy
to dissipate. The tape has minimal impact on the surface of the mirror
however, as it is adhesive if some of the corrosion is fragile it may be
taken away with the tape when removed, this is unavoidable but rarely
occurs. If any textiles are present the tape is not placed directly on it in
order to prevent losing any material. One potential alternative that has
not yet been explored is the use of carbon clips. These are non-adhesive
but would still provide a connection.
After the compositional data is collected, the polished edge is etched
to reveal the microstructure. For the mirrors in this investigation, the
etchant Ethanolic Ferric Chloride (EFC) (96 ml ethanol, 2 ml HCl and 5 g
FeCl
2
) was used as it had been shown to work well on copper alloys
(Scott, 1991:72), but others may be more suitable depending on the
metals present in the specimen. The etchant is applied to the polished
area using the ‘wipe’method where a cotton bud soaked in the etchant is
rolled over the surface of the metal until there is a sufcient etch. How
many swabs are required depends on a number of factors such as the
composition, as metal with a higher percentage of arsenic will be more
resistant to etching. Also, the extent of corrosion can impact the degree
of etch. The etchant preferentially reacts with copper so the solution
may react with the corrosion before the metal meaning further etching
may be required, but this is assessed on an individual basis. The main
advantage of using the ‘wipe’method is that the application of the
etchant is controlled and only exposes the necessary area of metal to the
solution. Once reacted the etchant is rinsed off using ultra-pure water
and neutralised using a 10 % Sodium Hydrogen Carbonate (NaHCO
3
)
solution to inhibit any further reaction with the material, it also mini-
mises the possibility of contamination by removing iron and chlorine.
The neutraliser is then washed off with ultra-pure water, ethanol and the
mirror is then dried. Once etched the mirror is placed back in the SEM
where micrographs of the microstructure are taken. The microstructure
is only observed via SEM as it is not possible to t a whole mirror placed
upright under an optical microscope.
5. Results
The four of the mirrors were manufactured using arsenical copper
ranging between 4.5 and 6 wt% As (Table 2). The average arsenic
content for the assemblage via a tangential taper section is 5.35 wt% and
5.21 wt% via a mounted ‘V’cross-section. Typically, the results from
either sampling technique fall within 0.5 wt% of each other. This is an
acceptable variation as the composition may vary throughout the sam-
ple and will change slightly with each repeated analysis. Figs. 10–13
show how the polished edge of the mirror in a tangential taper section
looks compared to the mounted ‘V’cross-section.
Fig. 10. Mirror E946. Left: TTS at 43x. Image width =2980 µm. Right: VS at 55x. Image width =2330 µm.
Table 2
Average bulk composition of the mirrors for both sampling methods.
Mirror Tangential Taper Section Mounted ‘V’Cross-Section
Elements (wt%) Elements (wt%)
Cu As Cu As
E946 94.26 5.74 94.56 5.44
E1452 94.27 5.73 94.02 5.98
E1519 94.72 5.28 95.14 4.86
E1541 95.36 4.64 95.42 4.58
E. Thomas and P. Gethin Journal of Archaeological Science: Reports 59 (2024) 104743
7

Fig. 11. Mirror E1452. Left: TTS at 170x. Image width =750 µm. Right: VS at 140x. Image width =920 µm.
Fig. 12. Mirror E1519. Left: TTS at 230x. Image width =560 µm. Right: VS at 120x. Image width =560 µm.
Fig. 13. Mirror E1541. Left: TTS at 100x. Image width =1280 µm. Right: VS at 75x. Image width =1710 µm.
E. Thomas and P. Gethin Journal of Archaeological Science: Reports 59 (2024) 104743
8

This alloy type and compositional range for Egyptian mirrors has
been seen in other studies (Eaton and McKerrell, 1976; Odler et al.,
2018; Rademakers et al., 2021 etc.) and would most likely but not
exclusively date them to the Middle Kingdom or earlier (Ben-Yosef,
2018).
Lead inclusions were identied in all three specimens alongside
other minor elements such as arsenic, iron, nickel and bismuth; these
impurities are common within Egyptian copper artefacts (Ogden, 2000:
152). Other iron sulphide inclusions were observed containing traces of
selenium and tellurium (Rehren and Northover, 1991); this type of in-
clusion has also been identied in other investigations (Kmoˇ
sek et al.,
2016 &2018, Rademakers et al., 2017).
The four mirrors displayed a recrystallized granular microstructure
(Figs. 14–17) indicating they have been through a number of cycles of
working once cast. The presence of twin lines shows that the metal was
annealed to relieve stress built up from hammering which allows for
further working. Slip lines caused by the stress of working are removed
during the annealing process (Scott, 1991: Fig. 12, pp.8) so their pres-
ence indicates that hammering was the nal stage of manufacture. This
type of microstructure has also been identied in Egyptian mirrors by
Odler and Kmoˇ
sek (2020) and Garenne-Marot (1984).
The micrographs (Figs. 18–21) have also enabled the identication
of segregation of arsenic in the mirrors, for both sampling methods.
However, due to the heavily corroded nature of E1519, the metal ap-
pears as patchy islands throughout the tangential taper section. The
original surface is no longer present or identiable. Where there are
arsenic-rich areas in the mounted ‘V’section for this specimen it is un-
clear if this was caused by corrosion or intentional enrichment.
Analyses have shown there were arsenic-rich γmetallic phases,
approximately 30 wt%. As on the surface of a number of specimens
(Fig. 22 and Table 3). The elemental map in Fig. 23 also demonstrates
how there was an increased quantity of arsenic around the outside of
sample E1452 as these maps show the distribution of the elements. The
purple corresponds with arsenic and green copper. The purple areas
show there is an increased quantity of arsenic compared to the bulk
metal underneath which appears greener. The arsenic-rich γmetallic
phases can also be seen as a distinct whiter band around the outside of
the samples, and the high arsenic interdendritic feeders can be seen
throughout E1452 and E1541. The arsenic-rich layer on the edge of the
sample would have produced a silver-coloured metal on the surface of
Fig. 14. Mirror E946. Left: TTS. Right: VS. Both at 500x.
Fig. 15. Mirror E1452. Left: TTS. Right: VS. Both at 500x.
E. Thomas and P. Gethin Journal of Archaeological Science: Reports 59 (2024) 104743
9

Fig. 16. Mirror E1519. Left: TTS. Right: VS. Both at 500x.
Fig. 17. Mirror E1541. Left: TTS. Right: VS. Both at 100x.
Fig. 18. Mirror E946. Left: TTS at 160x. Image Width =800 µm. Right: VS at 330x. Image Width =390 µm.
E. Thomas and P. Gethin Journal of Archaeological Science: Reports 59 (2024) 104743
10

the disk. It is still up for debate how such enriched layers were formed on
the surface, although the presence of interdendritic feeders in E1452 and
E1541 indicates the manipulation of the coring or inverse segregation
process rather than cementation that has previously been suggested
(M¨
odlinger and Sabatini, 2016: Tab.3, pp.72). It can also be noted that
corrosion had penetrated below the enriched layer deeper into the bulk
metal below, the gamma phase metal was more resistant to corrosion
(Northover, 1989).
6. Discussion
The purpose of this paper is not to provide an interpretation of
archaeologically relevant data related to diachronic technological
changes, but to demonstrate how this adapted minimally destructive
sampling and analytical method can be applied. For further discussion
regarding the data itself see (Thomas, 2024).
While a mounted ‘V’cross-section generally produces a larger
sampled area to examine and provides material from deeper within the
mirror compared to a tangential taper section, and although this may be
more useful in terms of metallurgical analyses as it is more probable to
Fig. 19. Mirror E1452. Left: TTS at 300x. Image Width =430 µm. Right:VTS at 430x. Image Width =300 µm.
Fig. 20. Mirror E1519. VS at 100x. No enrichment was identied in TTS..
Fig. 21. Mirror E1541. Left: TTS at 1000x. Image Width =130 µm. Right: VS at 550x. Image Width =230 µm.
E. Thomas and P. Gethin Journal of Archaeological Science: Reports 59 (2024) 104743
11

retrieve intact metal, it is unlikely to be approved by museum or con-
servation staff where there is a complete artefact due to its destruc-
tiveness. The types of data obtained from a tangential taper section
match that of a mounted ‘V’cross-section, but it is much less intrusive as
the sampling method exposes only the edge of the mirror disk compared
to a wedge being removed. Due to the nature of both sampling methods,
any issues identied for one will also be applicable to the other. This
gives a tangential taper section an advantage in terms of being much less
visually destructive which is benecial to museums more generally.
Additionally, a conservator could easily conceal such inconspicuous
damage introduced using this sampling method if required, which may
be more difcult to do with a ‘V’cross-section.
With this minimally destructive sampling technique, you gain in-
formation concerning composition and microstructure, allowing
Fig. 22. Location of analyses comparing enrichment with bulk metal. Corresponding chemical data can be seen in Table 3.
E. Thomas and P. Gethin Journal of Archaeological Science: Reports 59 (2024) 104743
12

research questions regarding their manufacture to be addressed, without
compromising the integrity of the whole artefact. It does not extremely
alter the appearance of the mirror, like with a ‘V’cross-section but gains
the same information as if one was taken. This means both the
researcher and museum gain valuable data on their artefacts without
impacting the overall ‘completeness’of the mirror or collection. As such,
a major advantage of this technique is that museums are more likely to
loan specimens out for research. This is primarily due to its minimal
destructiveness particularly when compared with asking to take a ‘V’
section. This means more mirrors (or other artefact types) can be ana-
lysed in greater quantities with the inclusion of not only composition but
microstructure, so in turn more data is collected and any trends, simi-
larities or differences between artefacts become easier to identify.
A benet of sampling with deep-led edge abrasion is that if the
mirrors were to later go on display within a museum, the sampling is
discreet enough that the tangential taper section is unlikely to be noticed
by the visitors, particularly if the mirrors are viewed behind glass from a
distance. This means important data can be obtained from the mirror
without compromising museums’displays or exhibitions. Equally, as all
data are shared with museums it would be possible to include some of
the research data, such as images of the microstructure, alongside the
artefact on display, which would enhance data dissemination to the
public.
Due to the sample location being on the side of the disk, not the face,
one advantage of this method is that it provides a cross-section through
the disk just like with a ‘V’cross-section. This means it is possible to
obtain a reasonable understanding of the internal material of the mirror
as well as enabling the identication of surface enrichment if it remains.
It is possible to gain microstructure data which in conjunction with
compositional information allows a fuller overall picture of the mirror to
be drawn, particularly compared to other analytical or sampling
methods like surface analysis or drill samples.
From the tangential taper sections, within at least three of the mir-
rors in this investigation, surface enrichment was identied and within
E1452 specically, the presence of interdendritic feeders demonstrates
that inverse segregation was most likely used to create a reective sur-
face. As previously stated, with surface analysis these kinds of micro-
structures would not be identiable and, furthermore, the ‘bulk’
composition of the mirror would be incorrect because of the surface
enrichment. Consequently, surface analysis alone cannot provide an
accurate bulk composition but can potentially indicate the presence of
surface enrichment if the amount of arsenic is higher than might be
expected, as was the case with a bracelet analysed by Rehren and Per-
nicka (2014) where pXRF showed 9.9 wt% arsenic on the surface when
the core sample only contained 1.2 wt%.
The primary limitations of this methodology are corrosion and
working with a small sample size. Corrosion can have a major impact on
the quality of the data gathered. Over time metals will corrode, the
extent of this varies but can impact the amount of metal left within an
artefact and can change the chemistry of the material through solid-state
diffusion. If the extent of corrosion is great that can leave minimal
amounts of metal to be examined which can have an impact on errors in
the data –this is most evident with microstructural interpretations of
mirror E1519 in this investigation. Moreover, the greater the extent of
corrosion the more material that needs to be removed during sampling
in order to reveal a sufcient amount of bulk metal. Also, over time
corrosion can completely replace the original metal within the mirror
meaning there may be no metal left to analyse (Gettens, 1961: 89). The
replacement of the metal with corrosion products over thousands of
years can result in the disk becoming fragile so the mirror may not be
suitable for sampling in the rst place.
Nevertheless, due to the magnifying capabilities of the SEM and
backscatter compositional mode, it is possible to easily identify differ-
ences in the material i.e. metal vs. corrosion. Areas of the sample that
have been greatly impacted by corrosion can therefore be avoided
during analyses. As such, it is possible to examine small sample sizes
comprising areas of intact metal. Being able to view the sample as an-
alyses are conducted reduces the impact of error and minimises the
uncertainty of the data. Additionally, for specimens where the presence
of corrosion is an issue additional analyses can be carried out in order to
verify the bulk composition. For example, instead of analysing three
large areas of metal over three sections of the sample, a larger amount of
smaller areas of intact metal avoiding corrosion are analysed before
being averaged out for the bulk composition.
Conversely, past conservation treatments such as stripping of
corrosion using acidic solutions were applied to a number of mirrors.
This removes the corrosion products to reveal the bulk metal underneath
(Jedrzejewska, 1964) however, this method can also remove the original
surface of the mirror. This is a potential issue when discussing surface
Table 3
Table showing a comparison in chemical data between surface enrichment and
bulk material.
Mirror Enrichment ¡1 (As wt%) Bulk Metal ¡2 (As wt%)
E946 30.01 5.37
E1452 29.64 5.92
E1519 28.10 5.01
Fig. 23. Elemental Mapping of surface enrichment in E1541. Cu =Green, As =Purple.
E. Thomas and P. Gethin Journal of Archaeological Science: Reports 59 (2024) 104743
13

enrichment and the original colour of the mirror as an enriched surface
may have been present but has since been removed and is no longer
observable. Although this is an issue it is not conned to this particular
sampling method as once corrosion is removed no analytical method
would be able to identify the original surface. On the other hand, mirrors
in this state require smaller quantities of metal for removal in order to
expose a sufcient area for analyses as there is little corrosion present.
This means they can still provide data regarding alloy type and
microstructure.
The main way in which uncertainty in the methodology and the data
being produced can be minimised is through comparison to ‘control’
specimens. Arsenical copper and tin bronze experimental pieces are
being produced as part of an ongoing doctoral project (Thomas). These
specimens are produced under known conditions i.e. cast only,
hammered and annealed etc. at relevant percentages. These are veried
using pXRF, SEM-EDX and MP-AES for their composition and SEM im-
aging for their microstructure. Comparisons between the exemplar
specimens and the archaeological data can be made to ensure the con-
clusions being drawn are viable. This reduces both the impact of error
and uncertainty in the archaeological data set.
The use of SEM-EDX can be considered another limitation of this
methodology. Firstly, access to an SEM-EDX and one with a large
chamber or stage tilting capabilities is not always possible for archae-
ological scientists. A large proportion of SEM chambers only accom-
modate small mounted samples hence the sizing is not suitable to t a
whole mirror. This means the institute in which the analysis is being
conducted must have a suitable SEM chamber and stage. Additionally,
there is the issue of transporting any specimens from the museums to the
institution where the SEM is housed, which can be difcult to organise
and facilitate. This means not everyone would be able to apply this
method to their material and would have to use alternative sampling
methods.
Secondly, the size of the SEM chamber restricts the investigation to
analyse mirrors smaller than 200 ×200 mm. This means a certain
proportion of mirrors are excluded, such as larger disks or mirrors with
handles still attached. It is important to achieve a representation
assemblage however due to the constraints of the equipment this be-
comes difcult and must be remembered when forming conclusions
about trends in the information. There will of course be some that are
larger than this but these are a minority, and often it is due to a handle
still being attached to the disk. For example, Manchester Museum have a
collection of 45 Egyptian mirrors. Based on their size, this technique
allows the analysis of 38 mirrors, equating to 84.5 %. The ones that
cannot be analysed are due to the fact that the handle is attached
technically making the mirror larger than allowed. Similar can be said
for the Garstang Museum, out of 36 mirrors 35 can be analysed, leaving
only one which again has a handle attached making it too big.
Provenance is an important aspect of all archaeological in-
vestigations as it places the object into broader cultural and social
contexts as well as aiding the identication of diachronic technological
changes. Unfortunately, the mirrors from the Garstang Museum have an
unknown provenance. This is partly due to the nature of the acquirement
of museum collections in the UK, for many artefacts, their context has
been lost or was never known. This can become an issue when trying to
link the information obtained to a wider socio-cultural framework
meaning any additions to the archaeological debate would be more
constructive knowing when and where the artefact came from. How-
ever, mirrors with no context can still be useful in building a reference
collection as comparative material, particularly in connection to how
surface enrichment was created. Future work adopting this sampling
method would benet from focusing on material with contextual in-
formation as with a larger assemblage of mirrors, with contextual in-
formation, there is the possibility of identifying trends chronologically
and geographically.
From the results provided above, deep-led edge abrasion producing
a tangential taper section analysed via SEM-EDX is a highly effective
method for identifying composition, segregation in the material or sur-
face enrichment and the microstructure of Egyptian metal mirrors.
Collectively the data gathered from analyses not only identies the alloy
type used but enables the mode of manufacture to be determined.
7. Conclusion
In this investigation the combination of the non-destructive analyt-
ical capabilities of SEM-EDX alongside being able to place the mirror as a
whole within the chamber enabling the collection of compositional and
microstructural data is ideal. The sampling method is minimally
destructive as only a small section of the surface has to be abraded in
order to gather sufcient information. Generally, this will not be visible
if the mirror is to be displayed. Although this paper focuses on only four
mirrors, this minimally destructive sampling and analytical methodol-
ogy was successfully applied to a large assemblage of 19 mirrors as part
of a Master’s thesis (Thomas, 2024). Further work is currently being
undertaken as part of a PhD project (Thomas) that will incorporate a
larger number of mirrors, data from which will become available upon
completion. This minimally destructive sampling and analytical meth-
odology demonstrates that a rather agreeable compromise can be met
between the analyst and the museum, where as little damage as possible
is inicted on the artefact but maximum information is gathered.
Funding
This research did not receive any specic grant from funding
agencies in the public, commercial, or not-for-prot sectors.
CRediT authorship contribution statement
Elizabeth Thomas: .Peter Gethin: Writing –review &editing,
Methodology, Data curation, Conceptualization.
Declaration of Competing Interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Data availability
No data was used for the research described in the article.
Acknowledgements
Firstly, we would like to thank the Garstang Museum at the Uni-
versity of Liverpool for granting access to the mirrors for research.
Secondly, we would like to thank Dr. Matthew Ponting and Naomi
Rubinstein for their insights.
References
Anheuser, K., France, P., 2002. Silver plating technology of the late 3rd century Roman
coinage. Historical Metallurgy 36 (1), 17–23.
Bayley, J., Crossley, D., Ponting, M., 2008. Metals and Metalworking: A research
framework for archaeometallurgy. Cumbria.
Ben-Yosef, E., 2018. Provenancing Egyptian metals: A methodological comment. Journal
of Archaeological Science 96, 208–215. https://doi.org/10.1016/j.jas.2018.06.001.
Casta˜
neda Reyes, J.C., 2010. Of Women, Mirrors and the Social Revolution
(Admonitions: 8, 5). G¨
ottinger Miszellen 225, 39–53.
Cope, L.H., 1974. The metallurgical development of the Roman Imperial coinage during
the rst ve centuries AD. Liverpool John Moores University (United Kingdom).
Cope, L.H., King, C.E., Northover, J.P., Clay, T., 1997. Metal analyses of Roman coins
minted under the Empire. British Museum, London.
Derriks, C. 2001. Les miroirs cariatides ´
egyptiens en bronze. M¨
AS 51. Mainz.
Eaton, E.R., McKerrell, H., 1976. Near Eastern alloying and some textual evidence for the
early use of arsenical copper. World Archaeology. 8 (2), 169–191. https://doi.org/
10.1080/00438243.1976.9979662.
E. Thomas and P. Gethin Journal of Archaeological Science: Reports 59 (2024) 104743
14

Garenne-Marot, L., 1984. Le Cuivre en Egypte pharaonique : sources et m´
etallurgie.
Pal´
eorient 10 (1), 97–126. https://doi.org/10.3406/paleo.1984.4352.
Gettens, R., 1961. MINERAL ALTERATION PRODUCTS ON ANCIENT METAL OBJECTS.
Studies in Conservation 6 (sup1), 89–92.
Gilmore, G.R., 1986. The composition of the Kahun metals. In: David, R.A. (Ed.), Science
in Egyptology. Manchester University Press, Manchester, pp. 447–462.
Gleba, M., Price, K., 2012. Textiles on Egyptian Mirrors: Pragmatics or Religion? In
Archaeological Textile Review. 2–13.
Goldstein, J.I., Newbury, D.E., Michael, J.R., Ritchie, N.W., Scott, J.H.J., Joy, D.C., 2017.
Scanning electron microscopy and X-ray microanalysis. Springer.
Jedrzejewska, H., 1964. The Conservation of Ancient Bronzes. Studies in Conservation 9
(1), 23–31. https://doi.org/10.2307/1505117.
King, C.E, and Northover, J.P. 1993. The Analyses, in: H. von Kaenel (Ed.), DerMünzhort
aus dem Gutshof in Neftenbach, Zürcher Denkmalpege Archaologische Monographien,
16, p. 110.
Kmoˇ
sek, J., Odler, M., Jamborov´
a, T., Msallamov´
a, ˇ
S., ˇ
S´
alkov´
a, T., Kmoníˇ
ckov´
a, M.,
2016. Archaeometallurgical study of copper alloy tools and model tools from the Old
Kingdom necropolis at Giza. Old Kingdom Copper Tools and Model Tools.
Archaeopress, Oxford, pp. 238–248.
Kmoˇ
sek, J., Odler, M., Fikrle, M., Kochergina, Y., 2018. Invisible connections. early
dynastic and old kingdom Egyptian metalwork in the Egyptian museum of leipzig
university. Journal of Archaeological Science 96, 191–207. https://doi.org/
10.1016/j.jas.2018.04.004.
Lilyquist, C., 1979. Ancient Egyptian Mirrors from the earliest Times through the Middle
Kingdom. Deutscher Kunstverlag, Berlin.
Masson-Berghoff, A., Pernicka, E., Hook, D., Meek, A., 2018. (Re)sources: Origins of
metals in Late Period Egypt. Journal of Archaeological Science: Reports 21, 318–339.
https://doi.org/10.1016/j.jasrep.2018.07.010.
Meeks, N., 1993. Surface characterization of tinned bronze, high-tin bronze, tinned iron
and arsenical bronze. In: Metal Plating and Patination. Butterworth-Heinemann,
pp. 247–275. https://doi.org/10.1016/b978-0-7506-1611-9.50025-x.
Michel, F. 1972. Analyse de quarante miroirs appurtenant au D´
epartement des Antiquit´
es
´
egyptiennes du Mus´
ee du Louvre. Annales 23 (Laboratoire de recherche des mus´
ees de
France): 34–46.
M¨
odlinger, M., Sabatini, B., 2016. A Re-evaluation of inverse segregation in prehistoric
As-Cu objects. Journal of Archaeological Science 74, 60–74. https://doi.org/
10.1016/j.jas.2016.08.005.
Northover, J.P., 1989. Properties and use of arsenic-copper alloys. Old World
Archaeometallurgy 111–118.
Odler, M., Kmoˇ
sek, J., 2020. Invisible Connections: An Archaeometallurgical Analysis of
the Bronze Age Metalwork from the Egyptian Museum of the University of Leipzig.
With Chapters by Katarína Arias, Veronika Dulíkov´
a, Lucie Jir´
askov´
a. Archaeopress
Egyptology 31. Archaeopress, Oxford.
Odler, M., Uhlir, K., Jentsch, M., Griesser, M., H¨
olzl, R., Engelhardt, I., 2018. Between
centre and periphery: Early egyptian and nubian copper alloy artefacts in the
collection of the kunsthistorisches museum vienna (KHM). ¨
Agypten Und Levante /
Egypt and the Levant 28, 419–456.
Odler, M., 2023. Copper in Ancient Egypt: before, during and after the Pyramid Age (c.
4000–1600 BC). Culture and History of the Ancient Near East.
Ogden, J., 2000. Metals. In: Nicholson, P., Shaw, I. (Eds.), Ancient Egyptian Materials
and Technology. Cambridge University Press, pp. 148–176.
Rademakers, F.W., Rehren, T., Pernicka, E., 2017. Copper for the Pharaoh: Identifying
multiple metal sources for Ramesses’workshops from bronze and crucible remains.
Journal of Archaeological Science 80, 50–73. https://doi.org/10.1016/j.
jas.2017.01.017.
Rademakers, F.W., Verly, G., Delvaux, L., Degryse, P., 2018. Copper for the afterlife in
Predynastic to Old Kingdom Egypt: Provenance characterization by chemical and
lead isotope analysis (RMAH collection, Belgium). Journal of Archaeological Science
96, 175–190. https://doi.org/10.1016/j.jas.2018.04.005.
Rademakers, F.W., Verly, G., Delvaux, L., Vanhaecke, F., Degryse, P., 2021. From desert
ores to middle kingdom copper: Elemental and lead isotope data from the RMAH
collection, belgium. Archaeological and Anthropological Sciences 13 (6), 100.
https://doi.org/10.1007/s12520-021-01329-w.
Rehren,T., and Northover, P. 1991. ‘Selenium and Tellurium in Ancient Copper Ingots’.
In Archaeometry, 90: [Proceedings of the 27th International Symposium on
Archaeometry, 2-6 April 1990, Heidelberg, Germany], edited by Günther A. Wagner
and Ernst Pernicka, 221–28. Basel: Birkh¨
auser.
Rehren, T., Pernicka, E., 2014. First data on the nature and origin of the metalwork from
Tell el-Farkha. The Nile Delta as a centre of cultural interactions between Upper
Egypt and the Southern Levant in the 4th millennium BC. Studies in African
Archaeology 13, 237–252.
Scott, D., 1991. Metallography and microstructure of ancient and historic metals. The J,
Paul Getty Trust.
Simpson, G., 1993. Analysis of ancient egyptian copper and bronze artifacts by using scanning
electron microscopy and metallographic techniques, M.Sc. dissertation, department of
archaeology. University of Liverpool.
Szpakowska, K., 2008. Daily life in ancient Egypt. Blackwell, Malden.
Thomas, E., 2024. Shining Light on Egyptian Mirrors: new scientic research into their
metallurgy. J. Archaeol. Sci.: Rep. 58, 104744.
Thomas, E., In Preparation. Reections in Time: investigating the social role of mirrors in
Ancient Egypt through metallurgical analysis. University of Liverpool [Unpublished
doctoral dissertation].
E. Thomas and P. Gethin Journal of Archaeological Science: Reports 59 (2024) 104743
15