ArticlePDF Available

Chemical composition and colouring agents of Roman mosaic and millefiori glass, studied by electron microprobe analysis and Raman microspectroscopy

Authors:

Abstract

About 100 fragments of Roman mosaic and millefiori glass were stylistically attributed to a Hellenistic type, a Ptolemaic and Romano-Egyptian period type and an early imperial period type. Twelve representative fragments were studied by electron microprobe analysis and Raman microspectroscopy. Eleven of them display a Na-pronounced recipe with low K, Mg and P contents, typical for the Roman period. Minor differences in composition are unsystematic, not reflecting the stylistic classification. Ionic colouring agents are Mn3+ for violet, Cu2+ for light blue, Co2+ for deep blue and Fe3+ for brown translucent colours. Calcium antimonates, lead antimonate and cuprite are the colourants responsible for white, yellow and red colours, respectively, and additionally serve as opacifiers. Mixing of ionic colouring agents and opacifying colourants led to a more differentiated palette of colours. Pb was used as yellow colouring agent, as a flux material and as a stabiliser for the colourant crystals. The remaining fragment consisting of a K-pronounced but still Na-bearing glass matrix was most likely produced during the Middle Ages or later.
ORIGINAL PAPER
Chemical composition and colouring agents of Roman
mosaic and millefiori glass, studied by electron microprobe
analysis and Raman microspectroscopy
V. Gedzevičiūtė & N. Welter & U. Schüssler & C. Weiss
Received: 9 September 2008 /Accepted: 19 February 2009 /Published online: 6 March 2009
#
Springer-Verlag 2009
Abstract About 100 fragments of Roman mosaic and
millefiori glass were stylistically attributed to a Hellenistic
type, a Ptolemaic and Romano-Egyptian period type and an
early imperial period type. Twelve representative fragments
were studied by electron microprobe analysis and Raman
microspectroscopy. Eleven of them display a Na-
pronounced recipe with low K, Mg and P contents, typical
for the Roman period. Minor differences in composition are
unsystematic, not reflecting the stylistic classification. Ionic
colouring agents are Mn
3+
for violet, Cu
2+
for light blue,
Co
2+
for deep blue and Fe
3+
for brown translucent colours.
Calcium antimonates, lead antimonate and cuprite are the
colourants responsible for white, yellow and red colours,
respectively, and additionally serve as opacifiers. Mixing of
ionic colouring agents and opacifying colourants led to a
more differentiated palette of colours. Pb was used as
yellow colouring agent, as a flux material and as a stabiliser
for the colourant crystals. The remaining fragment consist-
ing of a K-pronounced but still Na-bearing glass matrix was
most likely produced during the Middle Ages or later.
Keywords Roman glass
.
Mosaic glass
.
Millefiori glass
.
Raman microspectroscopy
.
Electron microprobe analysis
.
Ancient glass composition
Introduction
Mosaic glass, alternatively described as millefiori glass,
represents a very special kind of multifariously coloured
artistic glass and was used for the creation of exceptional
vessels as well as decoration plates for furniture and various
architectonical elements. In Mesopotamia, the earliest
manufacture of mosaic glass was substantiated for the
second millennium BC. It was the antique Romans,
however, who produced this kind of g lass for a broader
use. In the fifteenth and early sixteenth century this
technique was renewed by Venetian artisans and particu-
larly in the nineteenth century an almost exact imitation of
the antique techniques of mosaic glass m aking wa s
established in Venice, as well as in Bohemia and Silesia.
Objects of mosaic glass are composed of a multitude of
similar tiny glass sections or segments, which were cut off
from a long composite mosaic cane with a more or less
complicated and variously coloured pattern. The compo site
mosaic cane itself was previously made by thin rods of
differently coloured glass, which were then fused together
in a way to form the favoured combination of colours and
the desired pattern. To miniaturise the pattern by reducing
the diameter of the cane, the latter, while still hot, was
protracted and elongated with the pincers. The thin mosaic
cane was then cut crosswise to make small circular sections.
To form larger plates of mosaic glass, a number of sections
were arranged alongside one another and fused together
during a new round of heat treatment. In a further step,
these plates were sagged above concave or convex forms
Archaeol Anthropol Sci (2009) 1:1529
DOI 10.1007/s12520-009-0005-4
V. Gedzevičiūtė
:
C. Weiss
Institut für Altertumswissenschaften,
Residenzplatz 2,
97070 Würzburg, Germany
N. Welter
Institut für Physikalische Chemie, Am Hubland,
97074 Würzburg, Germany
U. Schüssler (*)
Lehrstuhl für Geodynamik und Geomaterialforschung,
Institut für Geographie, Am Hubland,
97074 Würzburg, Germany
e-mail: uli.schuessler@uni-wuerzburg.de
during another reheating procedure to create vessels, plates,
bottles, bowls or the like (Goldstein 1979; Grose 1989;
Stern and Schlick-Nolte 1994; Lierke 1999).
The results presented in t his paper are the first
concerning the composition of the glass recipes and the
character of the colouring agents used for the production of
Roman mosaic glass. A combination of Raman micro-
spectroscopy and electron microprobe analysis has turned
out to be a powerful tool for this purpose (Welter et al.
2007) and was applied to analyse representative pieces of
Roman mosaic glass.
Archaeology
The collection of antiquities in the Martin von Wagner
Museum in Würzburg holds about 260 fragments of ancient
mosaic glass which are generally attributed to a Roman
origin. The pieces were derived (1) from the comprehensive
art collection of Martin von Wagner, who was a painter and
sculptor, collector of antiquities and art agent of the
Bavarian King Ludwig I and (2) also from a collection of
Ludwig Brüls, a painter and friend of Martin von Wagners
(Möbius 1962). Most probably, Wagner and Brüls acquired
the mosaic glass fragments in Rome. This is supported by
the fact that most of the pieces are wrapped in cardboard
strips with gilded edges and were polished on their upper
surfaces, which was typical for ancient glass fragments
deriving from antiquity dealers in Rome. The museum
received both collections in 1858 and 1860, respectively.
Unfortunately nothing is known about the archaeological
sites were the fragments were found. Perhaps for this
reason, the pieces initially did not attract the attention of
archaeologists and were stored in the museums magazine.
In the course of a recent project, hundred of the mosaic
glass fragments were investigated regarding their stylistic
provenance (Gedzevičiūtė 2006). From this, a subdivision
of the fragments belonging to (1) a Hellenistic type, (2) a
Ptolemaic and Romano-Egyptian period type and (3) an
early imperial period type could be introduced. The earliest
fragments belong to the so-called Canosa-Group of the
Hellenistic period and are dated from the late third to the
late second century BC (Oliver 1968; Stern and Schlick-
Nolte 1994). Typical characteristics are spiral and radiant
patterns, but also small square monochrome segments
(tesserae) and segments of sandwich gold-glass (a piece of
gold leaf between two layers of colourless glass) which
were interspersed between mosaic glass sections. During
the imperial perio d from the middle of the first century BC
on, the older patterns are supplemented by new and very
variable kinds of ornamentation, by a remarkable increase
of translucent and opaque colours and by a wider variety of
vessel forms (Grose 1989). The radiant patterns disappear
and give place to more complicated, rosette- and flowerlike
(millefiori), stonelike and geometrical patterns. The Ptole-
maic and Romano-Egyptian period mosaic glass occurs
from the third century BC to the first century AD and
comprises plates and stripes as used e.g. for ornamentation
in architecture and inlays for furniture. This type of glass
points to a provenance from the Roman Egypt (Grose
1989).
From all hundred fragments, 12 pieces, which are repre-
sentative for the three groups described above, were selected
for analytical studies concerning the following questions:
1. Do the mosaic glass fragments in all their colours match
the very homogeneous typical Roman glass receipt?
2. Are all fragments of Rom an age, or can we find some
glass with older or younger glass compositions?
3. Do we find chemical differences between the three
stylistically different groups of the Roman mosaic glass?
4. What are the colouring agents of the translucent and the
opaque coloured glass?
Experimental
Samples
The representative selection of samples for analytical
studies is shown in Fig. 1. The fragments 2 and 6 were
attributed to the Hellenistic type; fragments 78, 83, 87 and
93 to the Ptolemaic and Romano-Egyptian perio d type and
fragments 7, 29, 33, 60 and 102 to the early imperial perio d
type. Fragment 103 was preliminarily classifi ed as Ptole-
maicearly imperial period type. The samples are com-
posed of differently coloured translucent or opaque glass
forming various designs and ornaments. All the fragments
are more or less intensively corroded on their lower side
due to long storage in soil. On their upper side, however,
the corrosion patina was removed and the glass was
repolished in the nineteenth century by previous owners
of the collections, so that a slight cleaning of the surface
with ethanol was sufficient to prepare the samples for
quantitative electron microprobe analysis and Raman
microspectroscopy. Subsequently, the matrix of all differ-
ently coloured glass parts and colouring agents of the
translucent glass were measured by electron-microprobe
analysis, whereas the Raman method was used to character-
ise the microcrystals contributing colour and opacity to the
various opaque glass parts.
Electron microprobe analysis
Chemical compositions of all the variously coloured glass
parts were determined at the Department of Geodynamics
16 Archaeol Anthropol Sci (2009) 1:1529
and Geomaterial Science at the University of Würzburg
using a CAMECA SX50 electron microprobe with wave-
length-dispersive spectrometers, allowing for non-destructive
in situ analysis of archaeological finds, if the following
requirements are met. The sample must be placed into the
microprobe as a who le. A special sample-holder was
constructed to allow the insertion of samples as large as
40×55×18 mm into the vacuum chamber. The cleaning of
the polished sample surface was performed with 100%
ethanol. To avoid carbon coating of the whole sample
surface, the fragments were enwrapped into aluminium foil
with small openings only at the surface locations selected
to be analysed. A conduction bridge was made using
conducting carbon lacquer, which may later easily be
dissolved in acetone. Oper ating conditions were 15 kV
accelerating voltage, 15 nA beam current and 25 μm beam
size. For glass samples such a large beam size has to be
used to avoid thermally induced diffusion of alkali elements
during analysis. Of course, the measured chemical compo-
sition reflects an integrated composition of the whole area
covered by the beam size; for opaque glass this means a
total analysis of glass matrix and colouring microcrystals
within the radiated area. Element peaks and backgrounds
were measured with counting times of 20 s each, but of
30 s for Fe, Mn, Cu and Co. For Sb analysis, the Sb Lβ
line was used instead of Lα to avoid interference with the
KKβ line; P was measured on the Kα line by using the
PET monochromator instead of the TAP to avoid interference
83
6
60
93
87
78
103
2
29
33
102
7
Fig. 1 Mosaic glasses investi-
gated. Fragments 2 and 6 are
attributed to the Hellenistic type,
fragments 78, 83, 87 and 93 to
the Ptolemaic and Romano-
Egyptian period type and frag-
ments 7, 29, 33, 60 and 102 to
the early imperial period type.
The size of the fragments is
3.2 cm longest side (2), 2.9 cm
longest side (6), 4.7 cm height
(7), 2.9 cm longest side (29),
2.8 cm longest side (33), 1.8 cm
longest diametre (60), 2.2 cm
side (78), 4.7 cm height (83),
2.6 cm longitude (87), 3.1 cm
longest diametre (102), 2 cm
height (103)
Archaeol Anthropol Sci (2009) 1:1529 17
with Ca Kβ second order line. Synthetic silicate and oxide
minerals or pure elements were used as reference standards. The
matrix correction was calculated by the PAP program supplied
by CAMECA. An analytical erroroflessthan1%relativefor
major elements is verified by repeated measurements on the
respective standards. For low element concentrations, the
analytical uncertainty increases. Using these operating con-
ditions, the detection limit is at about 0.030.1 wt.%.
From all 12 fragments, all differently coloured parts
were qualitatively analysed by wavelength-dispersive scans
and then quantitatively by about five single analysis runs.
From these, the average was calculated for the respective
glass part. Average compositions of the glass matrices of
the various glass samples are given in Table 1.
To visualise the differences in average atomic number
(i.e. general differences in chemical composition) between
variously coloured glass as well as the extremely fine-
grained crystals in the opaque glass parts by means of
backscattered electron images, the JEOL JXA-8200 elec-
tron microprobe at the Institute of Mineralogy at the
University of Erlangen-Nürnberg was used.
Raman microspectroscopy
All the differently coloured opaque parts of the glass
fragments were examined in a non-destructive fashion by
Raman microspectroscopy at the Institute for Physical
Chemistry of the University of Würzburg. The scattered
light was collected in backscattering geometry by focusing
a ×50 objective (Olympus ULWD MSPlan50) on the
entrance slit of a spectrometer LabRam, Dilor with
1,800 grooves/mm, respectively 950 grooves/mm diffrac-
tive grating. The 514.5 nm line of an argon ion laser
(Spectra Physics, Model 166) was applied for excitation.
The laser power at the focus spot with a beam diameter of
about 5 μm on the sample was kept below 5 mW. The
spectral resolution was 4 cm
1
. The detection system
consisted of a charge-coupled multichannel detector
(CCD, SDS 9000 Photometrics). The ac quisition of a
single spectrum typically took about 1060 s for the
crystalline phase; about ten accumulations were performed
for each Raman spect rum.
Analytical results
Principal glass composition
Apart from fragment 103, all glass fragments are composed
of a Na
2
O-rich sodium-calcium-silicate-glass, which is poor
in K
2
O, MgO and P
2
O
5
, but contains appreciable amounts
of CaO and also Al
2
O
3
. Beside the oxides of Si, Al, Ca, K
and Na, which define the basic composition of the glass
recipe, further components are added as colourants or as an
additional flux material. To get a common level for a
comparison of the basic compositions of the different glass
parts, all additional components were subtracted from the
analyses before normalizing the basic composition t o
100%. T his calculation procedure appears appropriate
assuming that Roman glass makers used o ne wid ely
distributed common recipe with these oxides to produce a
basical melt and not till then added the colourants and other
additional elements. If the fragments are Roman, the
calculated and normalised contents of the five oxides
should reflect the Roman basical glass r ecipe. The
variations of the normalised oxides for 11 fragments are
6977% for SiO
2
, 2.14.3% for Al
2
O
3
, 5.611.7% for CaO,
0.52.0% for K
2
O and 9.419.1% for Na
2
O (Fig. 2a).
Table 1 Average of microprobe analyses of every colour in every fragment
Fragment
(wt.%)
22 22 26 6 677 777292929
white yellow violet brown blue yellow white blue red yellow green white violet brown white brown
SiO
2
67.36 48.85 67.41 68.64 66.10 44.16 58.34 67.44 54.31 55.87 56.59 56.71 64.64 69.35 63.36 69.24
SnO
2
0.06 0.08 <0.05 0.07 0.39 0.05 0.03 n.d. 3.33 0.15 0.63 0.15 n.d. <0.05 <0.05 <0.05
Sb
2
O
3
5.60 1.72 0.12 0.03 <0.05 2.91 6.54 0.04 <0.05 1.44 3.72 6.29 0.15 <0.05 7.54 <0.05
Al
2
O
3
2.58 1.94 2.43 2.16 2.38 2.73 2.06 2.15 2.48 2.20 2.06 2.05 2.49 2.80 2.34 2.53
PbO <0.05 26.37 0.07 <0.05 1.62 30.39 12.46 0.99 9.41 16.41 10.27 12.33 0.67 0.00 <0.05 <0.05
CuO <0.05 0.62 0.11 <0.05 2.24 0.52 0.22 2.74 2.46 0.64 4.93 n.d. n.d. <0.05 <0.05 <0.05
CoO n.d. n.d. n.d. n.d. <0.03 n.d. n.d. <0.03 n.d. n.d. n.d. n.d. 0.03 n.d. n.d. n.d.
FeO 0.40 1.62 0.68 0.35 0.40 1.43 0.35 0.38 2.03 0.85 0.65 0.33 0.68 0.27 0.40 0.24
MnO <0.05 0.44 1.47 0.09 0.10 0.13 0.13 0.89 0.83 0.06 0.44 0.77 1.98 <0.05 1.30 0.07
CaO 7.37 5.11 8.02 8.79 7.71 7.35 6.37 7.13 7.62 5.71 5.64 6.51 7.72 7.99 8.08 7.77
MgO 0.41 0.35 0.54 0.55 0.46 0.42 0.63 0.53 0.66 0.34 0.43 0.55 0.59 0.55 0.65 0.46
K
2
O 0.65 0.58 0.76 0.78 0.75 0.59 0.62 0.95 0.88 0.75 0.67 0.76 0.88 1.25 0.75 0.70
Na
2
O 14.46 9.60 15.51 16.59 15.71 8.03 12.92 15.54 13.62 13.39 12.73 12.15 17.67 8.42 14.25 17.27
P
2
O
5
0.15 0.12 0.10 0.17 0.17 0.10 0.11 0.12 0.15 0.10 0.10 0.09 0.12 0.28 0.14 0.11
Cl 0.42 0.38 0.72 0.92 0.72 0.39 0.58 0.98 0.65 0.70 0.45 0.56 0.96 1.51 0.49 1.24
Total 99.47 97.77 97.94 99.12 98.75 99.17 101.37 99.88 98.43 98.58 99.31 99.25 98.57 92.42 99.29 99.63
33 33 33 33 33 60 60 78 78 78 78 78 83 83 83 83 83 83
white violet red yellow blue white blue white yellow violet red blue-
green
grey blue
heart
blue
centre
blue
stalk
violet yellow
n.d. not determined
18 Archaeol Anthropol Sci (2009) 1:1529
Fragment 103 displays a clearly different composition with
6771% for SiO
2
, 1.72.7 for Al
2
O
3
, 8.110.3% for CaO,
8.713.0% for K
2
O and 7.99.6% for Na
2
O (Fig. 2a). In
Fig. 2b, the average of all fragments except fragm ent 103
is compared with averages of Roman glass finds from
various pa rts of the empire, showing a very good
accordance and the extraordinary homogeneity of Roman
glass. Fragment 103, on its part (column 12), is compared
with different medieval K-rich glass compositions. Plotting
all single analyses of all glass samples into the diagrams
Na
2
O and P
2
O
5
versus K
2
O (Fig. 3) demonstrates that the
compositional range of the fragments covers the range of 45
Hellenistic and Roman glass samples, which were pub-
lished by Brill (1999). An exception is again fragment 103,
which clearly differs from all the others.
Ionic colouring agents
The colour of glass is in many cases caused by the addition
of certain minor- or trace-elements, which in their ionic
form are integrated into the network structure of the glass
matrix. This kind of glass is primarily translucent, as can be
seen in parts of the investigated samples. An additional
introduction of microcrystals as opacifiers and further
colouring agents leads to a formation of opaque glass and
to a mixture of colouring effects. For the glass samples
investigated, the following ionic colour ing agents could be
substantiated (see also Table 2):
Manganese: a purple colour is achieved by the addition
of Mn, which therefore mainly occurs in its trivalent
state when the atmosphere conditions in the glass-
furnace are oxidizing (Sellner et al. 1979). Mn contents
between 1.6 and 3.7 wt.% Mn
2
O
3
were detected in the
respective parts of samples 2, 7, 33, 78, 83 and 102.
Depending on the content of Mn
2
O
3
, but also on the
thickness of the glass, various shades of purple are
observed, reaching from pink to violet and to dark
purple with increasing Mn
2
O
3
-concentration: pink
1.7 wt.% in sample 2; violet 2.2 wt.% in sample 7;
dark purple 3.6 wt.% in sample 78. It should be noted
that these Mn
2
O
3
-values were calculated by stoichi-
ometry from Mn contents, assuming the major part of
Mn being trivalent as corroborated by the pink colours.
The effective proportions of MnOMn
2
O
3
MnO
2
in
the samples can be only achieved by direct spectro-
scopic investigations of the Mn oxidation state, which
was not carried out during the present study. As shown
by recent studies, however, the tetravalent Mn is not
really able to dissolve in a glass melt but resolves into
trivalent and divalent Mn (Kurzmann 2003). Therefore
it may not participate in a translucent glass as a
colouring agent. In its crystallised form, MnO
2
occurs
in an exceptional high average content of 7.9 wt.% in
the opaque dark part of sample 103 (see below).
Varying Mn-contents from the detection limit up to
1.5 wt.% MnO are also measured in glass samples without
purple, but different other colours. In these samples Mn
should occur in its divalent state as a consequence of a
more reducing furnace atmosphere. Thereby, Mn may
operate as a decolourizing agent to neutralise the colouring
effect of Fe (Sellner et al. 1979).
Cobalt: Co is responsible for a deep blue colour
(Wedepohl 2003) as observed in the samples 33, 60,
87, 102 and 103. It was detected in concentrations of
0.06, 0.09, 0.11 and 0.08 wt.% CoO, respectively, for
the first four samples. These low trace element contents
Table 1 (continued)
33 33 33 33 33 60 60 78 78 78 78 78 83 83 83 83 83 83
white violet red yellow blue white blue white yellow violet red blue-
green
grey blue
heart
blue
centre
blue
stalk
violet yellow
61.82 64.49 57.25 51.05 65.35 67.55 69.31 67.11 66.40 64.82 58.60 68.21 65.06 64.09 64.88 66.82 66.57 65.72
0.11 0.06 0.13 0.19 0.02 0.01 n.d. 0.08 0.05 n.d. 0.27 0.19 0.03 0.17 0.15 0.25 0.07 0.04
6.22 0.31 1.00 1.42 0.82 4.82 <0,05 3.19 1.20 0.46 0.26 3.11 0.24 0.33 0.39 0.57 0.29 0.62
2.32 2.49 2.81 2.37 2.62 2.57 2.60 2.01 1.97 2.24 2.25 2.08 2.33 2.03 2.26 2.02 2.05 1.98
5.85 0.84 10.71 23.15 2.06 <0.05 0.07 0.05 5.52 <0.05 6.88 0.12 0.08 1.25 1.51 2.88 0.20 5.68
<0.05 0.25 2.44 0.62 0.19 n.d. n.d. n.d. n.d. n.d. 1.55 1.95 0.33 1.36 1.71 1.70 0.36 0.27
n.d. n.d. n.d. n.d. 0.06 n.d. 0.09 n.d. n.d. <0.03 n.d. n.d. n.d. n.d. <0.03 n.d. n.d. n.d.
0.39 0.67 1.37 2.75 1.49 0.37 0.98 0.48 0.66 0.85 1.61 0.46 0.70 0.77 0.75 0.58 0.51 0.51
0.78 2.29 0.70 0.47 0.73 0.89 0.45 <0,05 0.08 3.16 0.30 0.04 1.89 0.16 0.16 0.08 2.18 0.07
6.74 7.79 7.30 5.27 7.55 7.55 7.07 6.83 5.90 7.15 8.81 5.66 9.39 9.89 8.81 6.69 7.74 7.36
0.47 0.58 0.65 0.37 0.49 0.61 0.43 1.61 0.49 1.29 2.08 0.46 0.99 1.26 1.11 0.64 0.84 0.62
0.71 0.91 0.88 0.72 0.85 0.73 0.71 0.54 0.47 0.96 1.70 0.51 0.63 1.15 1.10 0.53 0.73 0.67
14.72 17.64 13.16 10.13 16.64 15.05 17.14 18.06 16.70 17.42 13.10 17.93 15.95 15.03 15.85 17.63 16.95 15.80
0.08 0.16 0.12 0.07 0.12 0.17 0.08 0.07 0.06 0.26 0.69 0.08 0.12 0.37 0.34 0.09 0.19 0.05
0.64 0.94 0.69 0.42 0.99 0.45 1.18 1.04 1.18 1.01 0.75 1.19 1.04 0.93 1.00 1.17 1.15 1.02
100.84 99.41 99.19 99.00 99.97 100.75 100.11 101.07 100.67 99.62 98.84 101.99 98.75 98.76 100.01 101.64 99.81 100.38
Fragment
(wt.%)
83 83 83 87 87 87 93 93 102 102 102 102 103 103 103 103 103
yellow
stalk
yellow white white yellow blue yellow blue-
green
yellow white blue violet white dark red yellow-
green
blue
Archaeol Anthropol Sci (2009) 1:1529 19
are well in the position to give this deep blue colour to
the glass. A remarkably high value of 0.36 wt.% CoO
was measured in sample 103. The deep blue glass of
samples 33 and 103 contains additional opacifiers (Ca
antimonate, see below) which, however, does not really
influence the colour.
Copper: Cu in its divalent state was used to produce a
light blue colour in the translucent glass. The CuO
contents in samples 2, 6, 83 and 93 range between 1.4
and 2.7 wt.%. In all four samples, the translucen t light
blue colour was also used to evoke a very nice mixing
effect by underlaying part of the blue matrix with
yellow glass in spiral or radiant pattern (samples 2, 6),
as tiny yellow tubes (sample 93) or in figurative forms
(poppy flowers in sample 83: in this sample the blue
glass is also mixed with yellow pigments). In the
vicinity of the yellow parts, the light blue glass appears
in various green to blue-green colours, depending on
the thickness of the light blue overlayer (see Fig. 1).
In parts of the Cu-induced blue matrix of sample 83, Ca
antimonates are used as an opacifier. In sample 78, the Cu-
induced light blue is changed to an opaque cyan by addition
of Ca-antimonates. The opaque green colours in samples 7
and 103 are caused by a combination of CuO, PbO and
microcrystals (Ca antimonate in sample 7, cassiterite in
sample 103, see below). The samples contain 4.9 and
3.1 wt.% CuO and 10.3 and 24.4 wt.% PbO, respectively. It
is well known that Cu-induced blue glass with an addition
of considerable amounts of PbO turns to green (Weyl 1953;
Wedepohl 2003).
Iron: Fe in its trivalent state can be responsible for
the yellow-brown colour of glass (Scholze 1988;
Kurzmann 2003). In the translucent yellow-brown
parts of fragments 2 and 29, no colouring agents
except Fe were detected. Fe
2
O
3
contents of 0.39 and
0.29 wt.%, respectively, are therefore assumed to cause
this colour. Comparable or even higher Fe
2
O
3
contents
are detected in all the other coloured glass samples,
however, normally combined with decolo urizing Mn or
combined with other, more intensive colouring agents.
Within the yellow-brown glass of samples 2 and 29,
the MnO contents are exceptionally low, i.e. close to or
even below the detection limit of the microprobe, and
other colouring agents are missing. Therefore Fe
3+
can
evolve its typical yellow-brown tinge in the translucent
glass. In the vicinity of opaque white ornaments, the
brown turns to a honey-coloured yellow.
Microcrystals as opacifier and colourants
Looking at the backscattered electron images of the opaque
glass parts under the electron microprobe, it becomes
obvious that this kind of glass is intensively interspersed
by innumerable tiny crystallites, which play the role of
opacifying and colouring agents (Fig. 4). These are either
primarily supplied as crystal powder to the melt, or they
crystallise and grow inside the glass melt during the cooling
procedure, fed by certain major and/or trace elements of the
melt (Mass et al. 2002; Shortland 2002). For the first case,
an ideal moment for the addition of the crystal powder to
the melt has to be adjusted; this is done when the melt is
cool enough not to dissolve the crystals, but still viscous
enough to allow for mixing the crystals and the melt
(Shortland 2002). For the second case, depending on the
kind of crystals, a special heat-treatment of the melt during
Table 1 (continued)
Fragment
(wt.%)
83 83 83 87 87 87 93 93 102 102 102 102 103 103 103 103 103
yellow
stalk
yellow white white yellow blue yellow blue-
green
yellow white blue violet white dark red yellow-
green
blue
SiO
2
67.94 68.63 68.65 59.20 60.88 69.67 63.16 67.84 46.74 64.74 65.87 64.61 52.71 45.53 41.37 40.41 57.39
SnO
2
0.09 0.00 0.04 0.25 0.14 n.d. 0.04 0.09 0.07 0.06 n.d. n.d. 0.11 2.89 0.80 10.20 n.d.
Sb
2
O
3
0.70 1.02 0.67 9.40 2.10 0.76 0.86 0.29 2.32 5.58 0.03 <0.05 7.81 0.00 1.43 0.35 4.67
Al
2
O
3
2.27 2.27 2.17 2.19 2.34 2.16 1.99 2.10 1.87 2.22 2.26 2.16 1.74 1.27 1.67 1.55 1.40
PbO 3.78 0.01 0.70 8.18 10.49 n.d. 11.38 0.08 30.71 0.04 <0.05 0.10 7.89 15.44 25.95 24.42 2.10
CuO 0.20 <0.05 0.08 n.d. n.d. n.d. n.d. 1.69 n.d. n.d. n.d. n.d. n.d. n.d. 2.27 3.07 n.d.
CoO n.d. n.d. n.d. n.d. n.d. 0.11 n.d. n.d. n.d. n.d. 0.08 <0.03 n.d. n.d. n.d. n.d. 0.36
FeO 0.60 0.48 0.54 0.66 1.26 1.37 0.51 0.92 1.44 0.41 1.19 0.40 0.85 0.74 1.99 0.58 0.76
MnO 0.04 0.00 0.14 1.15 1.45 1.36 0.02 0.10 0.40 0.60 1.07 2.19 0.27 6.49 0.96 0.55 0.31
CaO 6.94 7.70 7.79 5.24 4.93 5.27 4.90 6.16 4.93 8.95 9.28 9.53 6.83 6.66 6.35 4.63 8.30
MgO 0.64 1.30 0.80 0.75 0.79 0.82 0.41 1.12 0.33 0.55 0.64 0.64 1.03 0.70 0.90 0.93 0.97
K
2
O 0.48 0.55 0.57 0.77 0.93 0.99 0.68 1.00 0.64 0.75 0.62 0.72 10.11 7.79 6.84 4.93 10.02
Na
2
O 16.64 17.32 16.81 12.15 12.06 16.39 14.63 17.71 8.09 15.76 16.45 16.91 6.11 6.49 5.38 5.42 7.40
P
2
O
5
0.04 <0.05 0.08 0.17 0.21 0.32 0.04 0.33 0.13 0.19 0.08 0.19 1.22 1.45 1.00 0.63 1.49
Cl 1.05 1.04 1.03 0.52 0.79 1.03 1.10 1.22 0.31 0.36 1.27 0.98 0.30 0.73 0.38 0.29 0.38
Total 101.36 100.31 100.05 100.61 98.33 100.25 99.73 100.65 97.98 100.21 98.83 98.44 97.00 96.18 97.30 97.96 95.55
20 Archaeol Anthropol Sci (2009) 1:1529
0
10
20
30
60
70
80
90
100
Na2O
K2O
CaO
Al2O3
SiO2
% normalized
2 6 7 29 33 60 78 83 87 93 102 103
0
10
20
30
60
70
80
90
100
12345678910 121314151617
Na2O
K2O
CaO
Al2O3
SiO2
% normalized
a
b
Fig. 2 a Basic glass composition including the oxides of Si, Al, Ca,
K, Na of all fragments, after subtracting the additional elements from
the analysis and normalisation to 100%. This procedure is carried out
to oppress the influence of the very variable quantities of additional
elements to the basic glass receipt, thus making the basic receipts
comparable with glass analyses from other occurrences, i.e. with
common Roman glass. b Comparison of the investigated fragments
with glass compositions from various Roman and Medieval prove-
nances: 1 average of all the investigated fragments, but without
fragment 103. 2 Average of 11 fragments of Roman cameo-glass of
the Martin von Wagner Museum Würzburg (Weiß and Schüssler
2000). 3 Average of five Roman cameo-glass vases exposed at the
British Museum London (Bimson and Freestone 1988). 4 Average of
36 Roman glass samples from Italy, Switzerland and Yugoslavia, first
third century AD (Braun 1983). 5 Average of 59 Roman glass samples
from Poitier, secondthird century AD (Velde and Gendron 1980).
6 Average of 48 Roman glass samples from Rouen, firstfourth
century AD (Velde and Sennequier 1985). 7 Average of 78 Roman
glasses from Cologne, firstfourth century AD (Rottländer 1990).
8 Average of 43 Roman glasses from Cosa near Rome, second century
BCthird century AD (Brill 1999). 9 Average of 20 Roman glass
samples from Aquileia, firstfifth century AD (Verità and Toninato
1990). 10 Average of 781 Roman glass samples from various parts of
the Roman Empire (Wedepohl 2003). 12 Fragment 103. 13 Average of
40 wood ash glasses of the Early Middleages from Germany, Norway,
France and England (Wedepohl 2003). 14 Average of 41 wood ash
glasses of the High and Late Middleages from Germany (Wedepohl
2003). 15 and 16 Knob beakers 84/9 and 84/27b, fifteenth century
AD, glassworks Nassachtal, Germany (Schüssler and Lang 2002).
17 Knob beaker, thirteenth century AD, Braunschweig (Bruckschen
2000)
Archaeol Anthropol Sci (2009) 1:1529 21
the cooling procedure may be necessary to initiate the
formation of seed crystals and to allow for crystal growing,
as described for example by Harding et al. (1989) for red
colourants. The size of the pigment crystals normally
ranges between less than 1 µm up to about 3 or 4 µm.
Occasionally, several crystals form small agglomerates of
about 10 µm. The number of crystals per volume differs
strongly from colourant to colourant and from sample to
sample.
To identify the tiny pigments, the spatial resolution of an
electron microprobe is not sufficient in many cases. Non-
destructive X-ray powder diffraction could be an appropri-
ate method, but in the case of the mosaic glass, a high
spatial resolution of the X-ray beam is essential, but not
yet standard for most of the instruments. Alternatively,
Raman microspectroscopy has turned out as a powerful
method, which works non-destructively, fast, with high
spatial resolution, and which allows for highly variable
sample sizes (Welter et al. 2007). For the studied glass
fragments, the following colourants and opacifiers have
been characterised:
Calcium antimonate is typical for all the opaque white
glass samples. Thereby a hexagonal CaSb
2
O
6
can be
distinguished from an orthorhombic Ca
2
Sb
2
O
7
. The
lack of a yet published Ca
2
Sb
2
O
7
spectrum led us to
synthesise both kinds of calcium antimonate to
produce suitable reference spectra. CaCO
3
and Sb
2
O
3
in a 1:1 molar ratio were mixed in a mortar and then
fired at 1,100°C for 10 h in a muffle furnace to receive
aCaSb
2
O
6
-dominated colourless powder. A molar
ratio of 2:1 at similar conditions yielded a Ca
2
Sb
2
O
7
-
dominated colourless powder. Repeated experiments
consistently resulted in mixtures of both calcium
antimonates, with respective predominance of one
type, however. The results were verified by X-ray
powder-diffractometry before taking the Raman refer-
ence spectra. As presented in Fig. 5a, hexagonal
CaSb
2
O
6
is c haracterised by Raman bands at 234,
323, 517 and 666 cm
1
, in good agreement with
published values (Husson et al. 1984). Orthorombic
Ca
2
Sb
2
O
7
displays characteristic Raman bands at 318,
367, 472, 624, 781 and 821 cm
1
(Fig. 5b). The white
glass of the investigated fragments is typically coloured
by the orthorhombic Ca
2
Sb
2
O
7
, i.e. in samples 7, 29,
33, 60, 78, 83, 87, 102 and 103. Figure 5d shows a
Raman spectrum typical for the white glass, taken from
sample 83. As an exception, white glass of sample 2
contains both CaSb
2
O
6
(Fig. 5c) and Ca
2
Sb
2
O
7
(similar
to Fig. 5d).
Calcium antimonates are also used as opacifiers for other
colours than white, but furtherm ore as colourant to produce
mixed c olours. In samples 33 and 103 the Co-induced blue
glass contains Ca
2
Sb
2
O
7
only as an opacifier. In sample 83,
the Cu-blue partly occurs opaque because of Ca
2
Sb
2
O
7
together with CaSb
2
O
6
. In sample 78, Cu-induced light
blue is turned to an opaque cyan by the addition of
CaSb
2
O
6
. In sample 7, the green colour of the matrix glass
is produced by Cu+Pb in combination with Ca
2
Sb
2
O
7
.
Lead antimonate Pb
2
Sb
2
O
7
, also called naples yel-
low, is the colouring and opacifying agent in all the
yellow glass parts investigated, i.e. in samples 2, 6, 7,
33, 78, 83, 87, 93 and 102. A representative spectrum
of fragm ent 87 is shown in Fig. 6, with characteristic
Raman bands at 142, 332, 450 and 506 cm
1
(Clark et
al. 1995). Interestingly, lead antimonate was also
detected in a blue-green part in the context of the
poppy flowers of sample 83. Most probably, the Cu-
induced light blue translucent matrix glass is doped by
a low amount of yellow lead antimonate to produce a
blue-green mixed colour (Fig. 4).
Cuprite Cu
2
O could be identified as the colouring
agent in all the red opaque glass samples, although the
crystallites with diametres between 0.5 and 1 µm are
clearly smaller, compared to the other crystal types
(Fig. 4 (7 red)). The characteristic Raman bands are at
154 and 216 cm
1
(Bouchard and Smith 2003), as
shown in Fig. 7a for the red glass of fragment 103. The
very irregular background of the cuprite spectra is a
5
10
15
20
25
wt.% K
2
O
wt.% Na
2
O
0.0
0.4
0.8
1.2
1.6
0 4 6 8 10 12 14
wt.% K
2
O
wt.% P
2
O
5
2
Fig. 3 Eleven fragments (filled squares) and fragment 103 (open
squares) in the diagrams Na
2
O and P
2
O
5
versus K
2
O, compared to 45
Hellenistic and Roman glasses (diamonds) (Brill 1999)
22 Archaeol Anthropol Sci (2009) 1:1529
Table 2 Compilation of colouring and opacifying agents for all differently coloured glass parts of all fragments
Ionic colouring agent Colouring crystals Opacifying crystals
Fragment 2
White CaSb
2
O
6
+Ca
2
Sb
2
O
7
CaSb
2
O
6
+Ca
2
Sb
2
O
7
Yellow Pb
2
Sb
2
O
7
Pb
2
Sb
2
O
7
Violet Mn
3+
Blue Cu
2+
Brown Fe
3+
Yellow brown Translucent brown overlaying opaque white
Green Translucent blue overlaying opaque yellow
Fragment 6
White Ca antimonate Ca antimonate
Yellow Pb
2
Sb
2
O
7
Pb
2
Sb
2
O
7
Blue Cu
2+
Green Translucent blue overlaying opaque yellow
Fragment 7
White Ca
2
Sb
2
O
7
Ca
2
Sb
2
O
7
Yellow Pb
2
Sb
2
O
7
Pb
2
Sb
2
O
7
Red Cu
2
OCu
2
O
Violet Mn
3+
Green Cu
2+
+Pb Ca
2
Sb
2
O
7
Fragment 29
White Ca
2
Sb
2
O
7
Ca
2
Sb
2
O
7
Brown Fe
3+
Yellow brown Translucent brown overlaying opaque white
Fragment 33
White Ca
2
Sb
2
O
7
Ca
2
Sb
2
O
7
Yellow Pb
2
Sb
2
O
7
Pb
2
Sb
2
O
7
Red Cu
2
OCu
2
O
Violet Mn
3+
Blue Co Ca
2
Sb
2
O
7
Fragment 60
White Ca
2
Sb
2
O
7
Ca
2
Sb
2
O
7
Blue Co
Fragment 78
White Ca
2
Sb
2
O
7
Ca
2
Sb
2
O
7
Yellow Pb
2
Sb
2
O
7
Pb
2
Sb
2
O
7
Red Cu
2
OCu
2
O
Violet Mn
3+
Cyan Cu
2+
CaSb
2
O
6
CaSb
2
O
6
Fragment 83
White Ca
2
Sb
2
O
7
Ca
2
Sb
2
O
7
Yellow Pb
2
Sb
2
O
7
Pb
2
Sb
2
O
7
Violet Mn
3+
Blue Cu
2+
±CaSb
2
O
6
±Ca
2
Sb
2
O
7
Blue-green Cu
2+
Pb
2
Sb
2
O
7
Pb
2
Sb
2
O
7
Fragment 87
White Ca
2
Sb
2
O
7
Ca
2
Sb
2
O
7
Yellow Pb
2
Sb
2
O
7
Pb
2
Sb
2
O
7
Blue Co
Fragment 93
Yellow Pb
2
Sb
2
O
7
Pb
2
Sb
2
O
7
Blue Cu
2+
Green Translucent blue overlaying opaque yellow
Fragment 102
White Ca
2
Sb
2
O
7
Ca
2
Sb
2
O
7
Yellow Pb
2
Sb
2
O
7
Pb
2
Sb
2
O
7
Violet Mn
3+
Blue Co
Fragment 103
White Ca
2
Sb
2
O
7
Ca
2
Sb
2
O
7
Red Cu
2
OCu
2
O
Blue Co Ca
2
Sb
2
O
7
Dark Mn
3+
MnO
2
MnO
2
Green yellow Cu
2+
+Pb SnO
2
SnO
2
Fragment 6: Ca antimonate was substantiated by electron microprobe, but not by Raman microspectroscopy
Archaeol Anthropol Sci (2009) 1:1529 23
7 red
83 yellow
29 white
white
brown
white
violet
yellow
red
green
yellow
turquoise
yellow greyish
blue
83
83
7
29
Fig. 4 View of selected sample parts under the electron microscope in
backscattered electron (bse) mode: a white helix of fragment 29 in a
brown glass matrix; the helix appears white also in the bse image
because of its high content of calcium-antimonate crystals which are
shown on the right side in a higher magnification. The brown glass
appears dark because of its lower average atomic number and the
missing of any crystals. A detail of fragment 7 shows the red core of a
flower surrounded by a yellow rim (with lead-antimonate as colouring
agent) and white leaves (with calcium-antimonate as colouring agent),
the latter embedded in a Mn
3+
-violet glass matrix without any crystals;
at the left side of the picture part of the green opaque glass
interstitially occurring between the flowers (with Pb and Cu as
colouring agent and calcium-antimonate as opacifier). The red core is
coloured by very finegrained cuprite crystals which are shown in
larger magnification at the right side. Fragment 83 with a heart-like
yellow detail of a poppy flower coloured by lead-antimonate crystals
as shown in larger magnification at the left (note that all three high
magnification pictures are displayed at the same magnification). The
torquoise is produced by a Cu-blue translucent glass, covering the
yellow opaque glass. The lowermost image of fragment 83 shows
the detail of a yellow stipe of a poppy flower, coloured by lead
antimonates, but inhomogeneously intermixed with the greyish blue
matrix glass to receive a somewhat diffuse appearance
24 Archaeol Anthropol Sci (2009) 1:1529
result of indistinct signals from the surrounding glass
matrix in consequence of very small grain sizes of the
cuprite and of fluorescence effects.
The formation of cuprite from the glass melt requires
accurate c alibration of the atmosphere in the melting
furnace. An oxidizing atmosphere produces divalent Cu
and therefore a light blue translucent glass. Strongly
reducing atmosphere leads to the formation of elementary
Cu crystals. Only under a slightly reducing atmosphere, the
monovalent Cu occurs to form Cu
2
O (Freestone 1987; Brill
and Cahill 1988). The size and the amount of the cuprite
crystals determine the shade of the re d colour. A n
increasing number of constantly small crystallites and an
increasing homogeneity of the cuprite distribution make the
colour of the glass more intensive and homogeneous. This
is influenced by the addition of lead to the glass melt
(Freestone 1987; Wedepohl 2003). Therefore, most of the
red opaque glasses described in literature and all of the red
samples presented here contain certain amounts of lead. An
increase of the crystal number and the homogeneity of their
distribution correlating to the concentration of lead was also
confirmed by backscattered images of the samples 7 and 78
(710 wt.% PbO) compared with sample 103 (26 wt.%
PbO). In the Raman spectra, this lead may be displayed by
a very typical band at 984 cm
1
(Colomban et al. 2003), as
shown in Fig. 7b, indicating that the glass matrix is formed
by a network of two-dimensional chains of SiO
4
tetrahedra
with mainly two non-binding oxygen atoms, in contrast to
Na- and K-dominated glass with mostly one n on-binding
b
c
200 300 400 500 600 700 800 900
fragment 83
white
Calcium-antimonate
(Ca
2
Sb
2
O
7
)
322
369
785
825
476
628
Raman Intensity
Wavenumber/cm
-1
200 300 400 500 600 700 800 900
239
325
340
521
670
fragment 2
white
Calcium-antimonate
(CaSb
2
O
6
)
Raman Intensity
Wavenumber/cm
-1
a
200 300 400 500 600 700 800 900
Calcium-antimonate
Ca
2
Sb
2
O
7
synthetic
318
367
821
781
624
472
Raman Intensity
Wavenumber/cm
-1
200 300 400 500 600 700 800 900
Calcium-antimonate
CaSb
2
O
6
synthetic
234
323
666
517
335
Raman Intensity
Wavenumber/cm
-1
d
Fig. 5 Raman-spectra of calcium antimonates; a synthetic CaSb
2
O
6
, b synthetic Ca
2
Sb
2
O
7
, c CaSb
2
O
6
as a white-colouring agent in fragment 2
and d Ca
2
Sb
2
O
7
as a white-colouring agent in fragment 83
Archaeol Anthropol Sci (2009) 1:1529 25
oxygen atom and a sheet-like network (Welter et al. 2007).
The growth of cuprite crystals is of course also depending
on temperature treatment (Henderson 1985).
Cassiterite SnO
2
serves as an opacifier only in the
green part of sample 103, which obtains the green
colour by a mixture of Cu and Pb. Within the spectrum,
the typical Raman bands are at 474, 633 and 775 cm
1
(Fig. 8) (Bouchard and Smith 2003). These bands are
almost identical to three bands of the calcium antim-
onate Ca
2
Sb
2
O
7
spectrum (318, 367, 472, 624, 781 and
821 cm
1
, see Fig. 5b, d). The spectra of both can be
distinguished, however, by the intensity proportions of
the specific bands and by the additional bands of
calcium antimonate.
Manganese oxide crystals, most probably γ-MnO
2
have
been detected in the Mn-rich dark centre of sample 103.
Characteristic Raman bands are at 243, 342, 384, 496,
562, 634 and 736 cm
1
(Fig. 9) (Julien et al. 2002).
Discussion and conclusions
As a result of the present study, the four questions asked at
the beginning may be answered as follows:
1. All the differently coloured parts of 11 of the
investigated fragments show the typical Roman glass
recipe, if additional elements are subtracted from the
analyses and the basical elements are normalised to
100%. This indicates that Roman glassmakers used a
standard glass recipe, whi ch was modified by the
addition of various colourants, opacifying agents and
further elements to produce certain special effects. The
fragments are formed by a sodium-calcium-silicate-
glass. Particularly, the low contents of K, Mg and P
substantiate the use of Na-minerals from Wadi Natrun
instead of plant ashes as a flux material, most indicative
for glass production in the Roman Empire. The
preferred flux mineral from Wadi Natrun is Trona
Na
3
H(CO
3
)
2
2H
2
O (Wedepohl 2003), but noteworthy
contents of Cl up to 1.3 wt.% in the glass matrix of the
fragments and a positive correlation of Cl and Na also
point to the use of NaCl in shares up to 2%, either
intentionally added or as impurity of the Trona.
2. Sample 103 differs clearly from all the other fragments
in several points. One major difference is the K-
pronounced composition of the basic recipe. K from
tree ashes as a flux material was introduced into glass
technology in the early Middle Ages (Wedepohl 2003).
200 400 600 800 1000 1200
332
450
506
142
fragment 87
yellow
Lead-antimonate (Pb
2
Sb
2
O
7
)
Raman Intensity
Wavenumber/cm
-1
Fig. 6 Raman-spectrum of lead antimonate as a yellow-colouring
agent in fragment 87
200 300 400 500 600 700 800 900 1000 1100 1200
216
154
fragment 103
red
Cuprite (Cu
2
O)
Raman Intensity
Wavenumber/cm
-1
200 300 400 500 600 700 800 900 1000 1100 1200
984
fragment 7
red
Cuprite (Cu
2
O)
154
Raman Intensity
Wavenumber/cm
-1
Fig. 7 Raman-spectrum of cuprite as a red-colouring agent in
fragment 103 and in fragment 7 together with the typical band for
lead glass at 984 cm
1
26 Archaeol Anthropol Sci (2009) 1:1529
Normally, however, glass containing tree ashes has low
contents of Na
2
O, in most cases below 2%, which is
not the case for fragment 103 wi th Na
2
O contents
around 6%. Nevertheless, such mixed glasses were also
produced during the Middle Ages, most probably by
recycling Na-rich and K-rich glass shards. KNa mixed
glass is reported from medieval glass factories, but also
from excavation finds (Bruckschen 2000 ; Schüssler and
Lang 2002; Brinker and Schüssler 2003). Concerning
the colours of fragment 103, the dark core is dominated
by Mn together with crystals of MnO
2
. This mineral
was not known as a colouring agent in antiquity, but
was used as a brown colourant during the Middle Ages,
i.e. for church windows and for glazing of apulian
ceramics (Clark et al. 1997; Wedepohl 2003). The
green part of the fragment contains cassiterite as
opacifier. This was not used in antiquity before the
second century AD (Turner and Rooksby 1961).
Cassiterite is a colourant, which from that time on
was commonly utilised to achieve white colours (so-
called tin-white). Interestingly, the white glass of
fragment 103 was not coloured by use of cassiterite,
but calcium antimonate. The PbO content of the red
opaque glass, 26 wt.%, exceeds clearly the PbO values
<10 wt.% in red glasses of the other fragments. The Co
content of the blue part is more than three times higher
than in the other blue glasses. Minor As contents
qualitatively detected in the blue glass of fragment 103
point to the use of Co from As-rich Co ore. Su ch ore
presumably derives from the As-bearing Co deposit of
Schneeberg in Germany which was earliest mined since
1520 (Geilmann 1962). From all these compositional
differences it becomes clear that fragment 103 does not
correspond to the other fragments. Fragment 103 is
assumed to derive from glass production of the Middle
Ages or even younger periods.
3. In terms of their styl istic provenance, no significant
differen ces could be recognised between the three
different types (1) Hellenistic type, (2) Ptolemaic and
Romano-Egyptian period type and (3) early imperial
period type, regarding the oxides of the basical recipe:
SiO
2
,Al
2
O
3
, CaO, K
2
O, Na
2
O, if normalised to 100%.
Within 1 σ standard deviation, the contents are similar
for the different stylistic types. Concerning the addi-
tional elements of the colourants and opacifyers, the
broad distribution of contents does not allow for a
chemical distinction of the stylistic types. Hereby, the
PbO contents of the yellow glasses between 16% and
30% in (1) and (2) and generally below 11% in (3)
form an exception. There is also a tendency to higher
PbO/Sb
2
O
3
ratios above 10 in (1) and (2) and to ratios
below 10 in (3). Cl contents >1% in the yellow and the
white glass parts of (3) tend to be enriched compared to
the contents <0.7% in (1) and (2). These weak differ-
ences may indicate a slight variation of the recip es with
time, region or glassmaking studio and, in case of Cl,
the use of more impure Trona for (3).
4. The colours of the mosaic glass are effected either by
addition of ionic agents or by colouring and opacifying
crystals or by a mixture of both. Thereby Mn
3+
is
responsible for purple, Co
2+
for deep blue, Cu
2+
for
light blue and Fe
3+
for yellow-brown translucent
colours, calcium antimonates for white opaque, lead
antimonate for yellow opaque and cuprite for red
opaque glass. The use of trivalent Mn and calcium
antimonate as colouring agents seem to be widespread
200 300 400 500 600 700 800 900 1000
633
775
474
fragment 103
green-yellow
Cassiterite (SnO
2
)
Raman Intensity
Wavenumber/cm
-1
Fig. 8 Raman-spectrum of cassiterite as opacifier in the green-yellow
parts of fragment 103
100 200 300 400 500 600 700 800 900 1000 1100 1200
736
247
384
342
562
496
634
fragment 103
black
Manganese oxide (MnO
2
)
Raman Intensity
Wavenumber/cm
-1
Fig. 9 Raman-spectrum of MnO
2
as a dark-colouring agent in
fragment 103
Archaeol Anthropol Sci (2009) 1:1529 27
in Roman glass production, i.e. the making of Roman
cameo glass (Weiß and Schüssler 2000).
Doping of translucent colours with calcium antimonates
produces opaque glass of more or less the same colour or,
alternatively, of a mixed colour, if high amounts of the
crystals are used. Various green colours are either produced
by stratification of translucent blue superimposed to opaque
yellow colours, or by mixing of translucent blue with lead-
antimonate, or by mixing of Cu and Pb, in cases with
additional colouring crystals.
The differently coloured glass parts of the 11 fragments
display unsystematically distributed Pb contents. For
example, half of the white glasses contain PbO at the
detection limit, the others show values between 0.7 and
12.5 wt.%. Beside Na, Pb was certainly used for further
reduction of the melting temperature. But Pb i s also
responsible for the yellow and, together with Cu, the green
colour of glass. It also acts as a stabiliser for colouring
crystals, as particularly shown for the red opaque glass
(Freestone 1987; Wedepohl 2003).
All the samples coloured by Cu contain minor amounts
of SnO
2
between 0.09 and 0.63 wt.%. This is taken as an
indicator for the use of bronze instead of pure copper as an
admixture to the glass melt. The composition of the bronze
may be recalculated, assuming Sn+Cu of the glass=100%
of the bronze: the Sn content of the suspec ted bronze was
14.8%, 11.3%, 4.9%, 8.8%, 14.5%, 10.5% and 4.9% for the
samples 2 (blue), 7 (green), 33 (red), 78 (cyan), 78 (red), 83
(blue) and 93 (blue), re spec tiv ely. The Sn-conte nt of
ancient bronze in general varies between 5% and 30%
(Blümner 1969). The red part of fragment 7 contains more
SnO
2
than CuO which does not match a bronze admixture.
The SnO
2
-content (10.2 wt.%) of the green glass in sample
103 exceeds by far the CuO content of 3.1 wt.%; in this
case further SnO
2
was intentionally added to produce the
cassiterite.
Acknowledgements Volker von Seckendorff managed the taking of
backscattered electron images at the Institute of Mineralogy, Erlangen
University. Klaus-Peter Kelber, Institute of Geography, Würzburg
University, took the pictures of Fig. 1. Ulrich Sinn and Irma
Wehgartner, Institute of Archaeology and Martin von Wagner
Museum, Würzburg University, facilitated the investigation of the
glass fragments and contributed helpful discussions. Simona Quartieri,
University of Messina, and an anonymous colleague carefully
reviewed the manuscript. Tobias Langenhan critically rea d the
manuscript and did some language correction. Thank you for all this
kind assistance.
References
Bimson M, Freestone IC (1988) An analytical studie of the
relationship between the Portland Vase and other Roman cameo
glasses. J Glass Studies 30:5564
Blümner H (1969) Technologie und Terminologie der Gewerbe und
Künste bei Griechen und Römern IV. Teubner, Leipzig
Bouchard M, Smith DC (2003) Catalogue of 45 reference Raman
spectra of minerals concerning research in art history or
archaeology, especially on corroded metals and coloured glass.
Spectrochim Acta 59A:22472266
Braun C (1983) Analysen von Gläsern aus der Hallstadtzeit mit einem
Exkurs über römische Fenstergläser. In: Frey O (ed) Glasperlen
der vorrömischen Eisenzeit I. Marburger Studien zur Vor-und
Frühgeschichte 5:129175
Brill RH (1999) Chemical analyses of early glasses, vol. II the tables.
The Corning Museum of Glass, Corning
Brill RH, Cahill ND (1988) A red opaque glass from Sardis and
some though ts on red opaques in general. J Glass Studies
30:1627
Brinker B, Schüssler U (2003) Spätmittelalterliche Glasproduktion im
Schönbuch bei Tübingen. In: Steppuhn S (ed) Glashütten im
Gespräch. Schmidt-Römhild, Lübek, pp 3945
Bruckschen M (2000) Faszinazion Glas. Die Braunschweiger Bod-
enfunde aus dem Mittelalter und der frühen Neuzeit. Dr Thesis,
University Kiel
Clark RJH, Cridland L, Kariuki BM, Harris KDM, Withnall RJ (1995)
Synthesis, structural characterization and Raman-spectroscopy
of the inorganic pigments lead-tin yellow type-I and type-II
and lead antimonate yellowtheir identification on medieval
paintings and manuscripts. J Chem Soc Dalton Trans 16:2577
2582
Clark RJH, Curri L, Henshaw GS (1997) Characterisation of brown-
black and blue pigments in glazed pottery fragments from Castel
Fiorentino (Foggia, Italy) by Raman microscopy, X-ray powder
diffractio n and X-ray photoelectron spectroscopy. J Raman
Spectrosc 28:105109
Colomban P, March G, Mazerolles L, Karmous T, Ayed N, Ennabli A,
Slim HJ (2003) Raman identification of materials used for
jewellery and mosaics in Ifriqiya. J Raman Spectrosc 34:205
213
Freestone IC (1987) Composition and microstructure of early opaque
red glass. In: Bimson M, Freestone IC (eds) Early vitreous
materials. Br Mus Occas Pap 56:173190
Gedzevičiūtė V (2006) Die Mosaikgläser des Martin von Wagner
Museums-archäologische und materialkundl iche Untersuchun-
gen. Magister Thesis, University Würzburg
Geilmann W (1962) Beiträge zur Kenntnis alter Gläser VII Kobalt als
Färbungsmittel. Glastech Ber 35:186192
Goldstein SM (1979) Pre-Roman and Early Roman glass in the Corning
Museum of Glass. Corning Museum of Glass, Corning, NY.
Grose DF (1989) The Toledo Museum of Artearly ancient glass.
Hudson Hills, New York
Harding RR, Hornytzkyi S, Date AR (1989) The composition of an
opaque red glass used by Fabergé. J Gemm 21:275287
Henderson J (1985) The raw materials of early glass production. Oxf J
Archaeol 4:267291
Husson E, Repelin Y, Vandenborre MT (1984) Spectres de vibration et
champ de force de lantimonate et de larséniate de calcium
CaSb
2
O
6
et CaAs
2
O
6
. Spectrochim Acta 40A:10171020
Julien C, Massot M, Rangan S, Lemal M, Guyomard DJ (2002) Study
of structural defects in γ-MnO
2
by Raman spectroscopy. J
Raman Spectrosc 33:223228
Kurzmann P (2003) Archäochemische Untersuchungen an mittelalter-
lichem Glas: Über farbloses Glas und Knochenasche-Glas. In:
Steppuhn S (ed) Glashütten im Gespräch. Schmidt-Römhild,
Lübek, pp 4649
Lierke R (1999) Antike Glastöpferei, ein vergessenes Kapitel der
Glasgeschichte. Phillip von Zabern, Mainz
Mass LJ, Wypyski MT, Stone RE (2002) Malkata and Lisht
glassmaking technologies: towards a specific link between
28 Archaeol Anthropol Sci (2009) 1:1529
second millennium BC metallurgists and glassmakers. Archaeo-
metry 44:6282
Möbius H (1962) Antike Kunstwerke aus dem Martin von Wagner
Museum. Erwerbungen 19451961. Stürtz AG, Würzburg
Oliver A (1968) Millefiori glass in classical antiquity. J Glass Studies
10:4870
Rottländer RCA (1990) Naturwissenschaftliche Untersuchungen zum
römischen Glas in Köln. Kölner Jb Vor-und Frühgeschichte
23:563581
Scholze H (1988) Glas: Natur, Struktur, Eigenschaften. Springer,
Berlin Heidelberg, NY
Schüssler U, Lang W (2002) Mineralogische Untersuchungen zu
Produktionseinrichtungen und Produkten der spätmittelalterli-
chen Glashütte Salzwiesen im Nassachtal bei Uhingen.
Hohenstaufen Helfenstein Hist Jb Kr Göppingen 12:3158
Sellner C, Oel HJ, Camara B (1979) Untersuchungen alter Gläser
(Waldglas) auf Zusammenhang von Zusammensetzung, Farbe
und Schmelzatmosphäre mit der Elektronenspektroskopie und
der Elektronenspinresonanz (ESR). Glastech Ber 52:255264
Shortland AJ (2002) The use and origin of antimonate colorants in
early Egyptian glass. Archaeometry 44:517530
Stern EM, Schlick-Nolte B (1994) Frühes Glas der alten Welt.
Sammlung Ernesto Wolf. Hatje, Stuttgart
Turner WES, Rooksby HP (1961) Further historical studies based on
X-ray diffraction methods of the reagents employed in making
opal and opaque glasses. Jb RGZM 8:16
Velde B, Gendron C (1980) Chemical composition of some Gallo-
Roman glass fragments from Central Western France. Archaeo-
metry 22:183187
Velde B, Senne quier G (1985) Observations on the chemical
composition of several types of Gallo-Roman and Frankish glass
production. Annales 9e Congrès de lAIHV, 127147
Verità M, Toninato T (1990) Riscontri analitici sulle origin i della
vetraria Veneziana. Contributi Storico-Tecnici 1. Comitato
Nazionale Italiano AIHV
Wedepohl KH (2003) Glas in der Antike und im Mittelalter.
Geschichte eines Werkstoffs. Schweizerbart, Stuttgart
Weiß C, Schüssler U (2000) Kameoglasfragmente im Martin
von Wagner Museum der Universität Würzburg und im
Allard Pierson Museum Amsterdam. Jb Dt Arch Inst
115:199251
Welter N, Schüssler U, Kiefer WJ (2007) Characterisation of inorganic
pigments in ancient glass beads by means of Raman micro-
spectroscopy, microprobe analysis and X-ray diffractometry. J
Raman Spectrosc 38:113121
Weyl WA (1953) Coloured glass. Soc Glass Technology, Sheffield
Archaeol Anthropol Sci (2009) 1:1529 29
... Raman spectroscopy is useful for analysing glass colourants and opacifiers particularly when colouring elements and crystals are present in very small concentrations. The technique is rapid, non-destructive and has a high spatial resolution which allow for a range of sample sizes to be studied (Gedzeviciute et al., 2009). The technique has also been used to study glass corrosion processes (Robinet et al., 2005). ...
... Sodium and lighter elements can be under reported as they are mobilised by the x-ray beam and relatively large beam sizes for glass samples are required in order to avoid thermally induced diffusion of the alkali components (Gedzeviciute et al., 2009). The use of low vacuum SEM and lower x-ray excitation energy can also help to reduce this. ...
Thesis
Full-text available
In comparison to the rest of Britain and Europe, there has been relatively little research investigating Scottish medieval and post-medieval window glass. It is thought that all glass was imported into Scotland prior to the first known glass furnace being established in 1610 AD. There is limited documentary evidence for the importation of window glass prior to this although there has been more substantial examination of the documentary evidence for the Scottish glass industry in the Post Medieval period. Minimal scientific work to corroborate the documentary evidence for window glass importation and production had been previously carried out. Evidence for window glass in Scotland during the medieval and post medieval period is reviewed and placed in the wider context of window glass production in Europe. A summary of the evidence of window glass production in Scotland up to the first known excavated glasshouse in Scotland is also presented. Previous scientific analysis of medieval window glass from European contexts have been reassessed and a model to provenance location manufacture, based on major elemental composition and rare earth element profiles, is proposed. Glass from the post medieval period is compared to an existing model proposed for English window glass manufacture. Window glass found in Scottish contexts from the 12th – 18th centuries was chemically characterised to shed light on the origin of window glass used in Scotland. Two hundred and forty-nine samples of medieval window glass and two hundred and two samples of post-medieval window glass fragments were analysed by portable-X-Ray Fluorescence (p-XRF), Scanning Electron Microscopy–Energy Dispersive X-ray Spectroscopy (SEMEDS) and fifty nine of these samples were analysed by Laser-ablated Inductively Coupled Mass Spectroscopy (LA-ICP-MS). The main type of glass in the 13th century appears to have been a potassium rich glass made in North Western France (Normandy). However, there was an increase in the use of high lime low alkali (HLLA) glass in Scotland as early as the 14th century. This glass type was likely made in North-West Germany/Rhineland or in the Lorraine/Argonne region of France. Later monastic sites continued to use this cheaper HLLA glass which was most likely imported into Perth from Bruges and Antwerp in the late 14th and 15th Centuries. Coloured glass was still predominantly made in the North West France, although blue coloured glasses were made in a variety of locations using different baseglass and colourant recipes. The composition of post medieval window glass primarily follows the English model from the start of the 17th century. Imported HLLA 1 glass is replaced by a HLLA 2 glass manufactured in Britain. The level of strontium increases in HLLA glass during the 17th century with levels higher than seen in similar glass types made in English furnaces, suggesting that kelp was increasingly used during this period as a flux and may be a sign of Scottish manufacture. By the start of the 18th century, a mixed alkali type glass was being made fluxed with kelp.This work therefore represents the largest analytical study of Scottish medieval and postmedieval window glass undertaken and presents insights into the transfer of materials, technologies and trade routes during the period as well as the early glass manufacturing industry in Scotland Available at https://www.ros.hw.ac.uk/bitstream/handle/10399/4405/SpencerH_0620_egisSS.pdf?sequence=1&isAllowed=y
... As already noted, turquoise glasses are opacified through Ca antimonate inclusions, often characterized by anhedral habit and minor substitutions of Ca with Na and, more frequently, low amounts of Pb (when present, it usually accounts for 2-4 wt%). This is consistent with what is reported in the literature on Ptolemaic (Bimson & Freestone, 1988) and Roman glass (Basso et al., 2014;Gedzeviči utė et al., 2009;Silvestri et al., 2014) glasses were also reported, but mainly dated to the Late Antiquity. Lead antimonate, in association with copper, is also known as the main colorant/opacifier in the few analyzed LBA opaque green glasses from Egypt (Shortland & Eremin, 2006), but has not been identified in the Tebtynis samples analyzed during this work. ...
Article
Full-text available
Inlays range among the most aesthetically pleasing and technically challenging glasses produced in the Ptolemaic period. Despite the central role of this phase in the history of glass technology, little is known about the recipes and the technological knowledge of the Egyptian artisans. This paper will thus focus on the study of the materials from the secondary workshop of Tebtynis (Fayum oasis, Egypt). We report the first multi‐methodological study comprising textural, chemical, and mineralogical analyses (OM, SEM‐EDS, EPMA, and μ‐Raman) on a set of 81 colorless, white, blue, and green samples carefully selected among the 800+ glasses from the craft area now stored at the Museo Egizio, Torino (Italy). Our study offers the biggest compositional database of well‐dated Ptolemaic glasses currently available in the literature, highlighting some interesting novelties regarding the silica and alkali sources, and the coloring and opacifying techniques employed. The results suggest a specialized craft of traditional origin, but open to innovation and experimentation, as expected from transitional phases.
... Based on the results of the ICP-MS analysis, though, we hypothesise that the colouring agent in the pink powder was Fe found in concentrations of 0.50 mg·g −1 , probably in the form of hematite. Copper, e.g., in the form of cuprite or copper nanoparticles, was also used as a red colourant in historical glassmaking [32,33], but the amount of Cu used in such cases was about 100x greater than in the red powder from Prostějov (0.26 mg·g −1 , Table 2). Py-GC/MS analysis revealed the presence of organic additives in the pink faience powder, i.e., diterpene resin acids of the abietane and pimarane types (Figure 3a). ...
Article
Full-text available
An archaeological excavation in Prostějov (Czech Republic) revealed a workshop of a local potter with colourless, pink, and blue powders presumably used to produce faience/surface decoration. A comprehensive analytical study, which combined elemental and molecular analysis techniques, was performed to shed light on the chemical composition of these unique findings. Scanning electron microscopy with energy dispersive X-ray spectroscopy (SEM EDX), inductively coupled-plasma mass spectrometry (ICP MS), flow injection analysis (FIA) with electrospray ionisation mass spectrometry (ESI MS), laser desorption ionisation mass spectrometry (LDI MS), and Raman spectroscopy were applied to reveal the elemental composition of the powders and identify the colouring agents in the pink and blue powders. The colouring agents in the pink powder were probably iron and the agent in the blue powder is Prussian blue. On top of that, it was also possible to determine the organic additives in these powders through pyrolysis gas chromatography with mass spectrometric detection (Py GC/MS), atmospheric solids analysis probe ion mobility mass spectrometry (ASAP IM MS), and LDI MS. The organic constituents were identified as plant resin, beeswax, and fats. These results point to the preparation of faience/pigment mixtures as oil paint.
... The Raman spectra of these two samples are very similar (Fig. 5B). The sharp peaks with a medium intensity located at 670 cm −1 , the broad peak with a lower intensity at 234 cm −1 , and the broad low-intensity peak at 330 cm −1 can all be attributed to hexagonal Ca-antimonate (CaSb 2 O 6 ) in agreement with the values published by Gedzevičiūtė et al. [62] and Husson et al. [63]. The very low intensity peaks usually located near 520 cm −1 is not visible in the present spectra, maybe because of the low crystallinity of the Ca-antimonate, or maybe because it is hidden by the broad bands of the amorphous glass [64]. ...
Article
Full-text available
An excavation in 1856 by Charles T. Newton and a re-excavation in 1990–93 by a joint Danish-Turkish team revealed several mosaic floors in a late-antique domus from the fifth century CE, now called the House of Charidemos. Nineteen tesserae from the floor have been analysed by laser ablation inductively coupled plasma mass spectrometry, Raman micro-spectroscopy, scanning electron microscopy with energy dispersive X-ray analysis, and X-ray diffraction. Seven tesserae were made of opaque glass, eleven from various rock/lithic materials, while one of them was a ceramic fragment. This case-study reports the first analyses undertaken of tesserae from late-antique Halikarnassos. The results show the use of recycled Sb–Mn decoloured glass and two types of red glass. A comparison with tesserae from other sites in Anatolia from the same period shows similarities in the base glass composition, but also some particularities of the colouring and opacifying agents used for the mosaic of the House of Charidemos. The characterisation of the stone tesserae shows a homogeneity in the choice of the materials of the same colours, but no certain provenance has been established in the present work mainly due to the lack of comparative materials. Graphical Abstract
... Both tin and arsenic were found to be associated to copper in the turquoise glass of Vinha das Caliças 4 (Fig. 5). The presence of tin in glass is generally associated to the use of tin-rich bronze scrapings as the copper source, as reported in Egyptian, Phoenician-Punic, and Roman glass (e.g., Gedzevičiūtė et al. 2009;Shortland and Tite 2000;Truffa Giachet et al. 2019). Arsenic commonly occurs in both copper and lead ores, from which it is primarily exploited (Adriano 2001), and has been found in low concentrations (ca. ...
Article
Full-text available
Phoenician-Punic glass beads recovered from the necropolis of Vinha das Caliças 4 (Beja, Portugal), dated to the Iron Age, were studied by variable pressure scanning electron microscopy coupled with energy dispersive X-ray spectroscopy (VP-SEM–EDS) and laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS). All glass was found to be soda-lime-silica natron-based glass. Trace element and rare earth element analysis suggest that the blue, turquoise, white, and colorless glass was manufactured in the Levant region, using rather pure coastal sands, while two amber and black-colored glass beads seem to have been produced in Egypt. Although the exact manufacture location of these Phoenician-Punic glass beads remains unknown, these results are in agreement with the commercialization of raw glass produced in established primary glass workshops in the Levant and in Egypt. The procurement of cobalt ores from specific geological provinces, on the other hand, is a strong evidence of the existence of large-scale trade routes of both raw materials and finished artefacts in the Iron Age. The finished artefacts trade routes included not only the previously established Mediterranean world and sub-Saharan Africa, but also Southern Central Iberia.
Chapter
The basis of the generation of characteristic Raman spectral signatures of glass, pottery and enamelled artefacts is considered and the theory underlying the structural changes taking place in glassy and crystalline silicates. A historical survey is undertaken of glass compositional changes through the ages and the Raman spectral fingerprints that are available. The resonance Raman effect and its application to glass structural studies. Glass weathering mechanisms and their correlation with Raman spectral signatures. Use of mobile Raman spectroscopy. The recognition of technological milestones in glass production through eth ages. Case studies of coloured glasses and trade glass beads.KeywordsGlassPotteryEnamelsGlass trade beadsRaman spectral signatures of glassWeathering of glassBlueGreen and yellow coloured glasses
Article
Full-text available
The Great Mosque of Córdoba is the only mediaeval Islamic monument in al-Andalus that still contains some of its original 10th-century glass mosaic decoration, even though they were significantly altered and restored several times in the 18th, 19th and 20th centuries. By examining analytical, textual, and graphic evidence, this study attempts to reconstruct the different transformations of the Maqṣūrah, focusing on the glass mosaics in order to assess possible implications for their state of preservation. The results show that high-alkaline glass, which can be associated with modern tesserae, was used to restore the Miḥrāb and Sābāṭ mosaics. Modern pigments were also detected in the paints that were applied in situ to the surface of the colourless glasses. Textual and analytical evidences prove that Bāb Bayt al-Māl mosaic was completely removed in the 20th century and a replica with high-lead tesserae was installed in the same place.
Article
Mobile Raman and XRF systems are particularly suitable for the study of cultural heritage objects, allowing in situ, nondestructive analysis on a large variety of materials. Nowadays, thanks to technological developments, the size of the new generation of mobile instruments is reduced. Moreover, operation of the instrumentation is straightforward, as most instrumental settings and data interpretation are automated. However, automated data processing should be evaluated carefully and the risks of incorrect data collection and/or interpretation must be considered and prevented. Therefore, the aim of this research is to start a discussion about some of the challenging aspects and pitfalls, which can be encountered when using mobile techniques. Data collection and interpretation are examined when dealing with works of art. Those aspects are highlighted based on a measuring campaign carried out in Tuscany (Italy) in three different contexts: an archaeological excavation in the Late Roman Villa dell’Oratorio (Capraia-Limite sull’Arno, Florence), the analysis of a mosaic at the façade of the Pisa Cathedral and studies inside the Museo dell’Opera del Duomo (Pisa), where some outdoor statues are exhibited. Measurements were acquired with two state-of-the-art mobile Raman and one handheld X-Ray fluorescence (XRF) systems.
Article
This paper provides an overview of an important glass decorative technique of the Renaissance: the pick-up decoration, including millefiori and splashed technique. Different subjects of investigation are linked: I) history and nomenclature, II) production technique, III) a survey of literature on archaeological findings, and IV) chemical composition. The results of an exhaustive survey on available data concerning the archaeological contexts where glass fragments decorated with this technique were found are presented, in order to establish the possibility of the use of decorative patterns as an indicator of provenance. This work also proposes a uniformed nomenclature to be applied to pick-up techniques. Comparing the artefacts that were found in archaeological contexts with the objects displayed in museums one can note that the first ones do not follow the same trend than the last, either in forms or in colours’ selection. So far, Portugal is the country with more studied archaeological contexts where glass fragments decorated with this technique were uncovered and different and unique patterns for murrine were identified. This work brings new and exciting insights about the production and trading of glass with pick-up decoration in Europe during the Renaissance period.
Article
Full-text available
Elemental analyses have been conducted on 61 coloured opaque glasses from the Malkata and Lisht New Kingdom glass factories. The presence of tin in several of the blue glasses suggests that a bronze casting byproduct or corrosion product was the source of the copper colorant for these glasses. A positive correlation between the lead and antimony concentrations of the yellow and green opaque glasses, plus a consistent excess of lead oxide in these glasses, suggests the use of antimony-rich cupellation litharge as the source for the Pb2Sb2O7 colorant in these glasses. The metallurgical byproducts used to colour the Malkata and Lisht glasses provide an explicit mechanism for Peltenburg's theory of interaction between second millennium BC glassmakers and contemporary metalworkers.
Book
Das Buch erörtert zuerst die besondere Natur des Glases, indem - dem Herstellungsprozeß folgend - zunächst das Verhalten von Glasschmelzen betrachtet wird, um daraus die Struktur der festen Gläser abzuleiten. Je nach chemischer Zusammensetzung ergeben sich dafür charakteristische Eigenschaften, besonders bei den nichtsilicatischen, nichtoxidischen und metallischen Gläsern. Die zweite Hälfte des Buches befaßt sich mit den Eigenschaften der Gläser, die nach Möglichkeit aus den Strukturen abgeleitet werden. Dabei wird auch auf die Meßmethoden und die Einflüsse von Zusammensetzung, Temperatur und Vorgeschichte eingegangen, wobei die neuesten Fortschritte in der Entwicklung von Gläsern mit höherer Festigkeit und besseren optischen, elektrischen und chemischen Eigenschaften behandelt werden. Neue Abschnitte sind auch der Glasoberfläche und dem Sol-Gel-Prozeß gewidmet. Das Buch ist in seiner Anlage ohne Konkurrenz im deutschen Sprachbereich. Es stellt nicht nur eine Einführung in den Werkstoff Glas für Lernende dar, sondern ist auch durch viele praktische Hinweise ein wertvolles Hilfsmittel bei der Anwendung von Glas. Sehr viele Literaturzitate ermöglichen einen schnellen Zugriff zu ausführlicheren Quellen. Die vielseitigen und oft einzigartigen Eigenschaften von Gläsern werden aus der Glasstruktur abgeleitet, die eingehend behandelt wird. Daraus ergibt sich das Verständnis für die Einflüsse von Zusammensetzung, Temperatur und Vorgeschichte. Das Buch ist nicht nur ein Lehrbuch, sondern auch ein Hilfsmittel für den praktischen Gebrauch von Glas.
Article
The origin of the colorants in Egyptian glass of the second millennium BC has been the subject of much research and debate. Several colorants including lead antimonite yellow and calcium antimonite white appear in the archaeological records apparently concurrently with the introduction of glass, and its is possible that their origins are in some way linked. This paper examines the use of the antimonite colorants and uses analytical and experimental techniques to decude possible technologies of production. Trace element data derived from a pilot study by LA_ICPMS gives additional indications of a possible source in the Caucasus for the antimony of these glasses.