See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/257154732

Egyptian faience glazing by the cementation

method part 1: an investigation of the glazing
powder composition and glazing ....

Article in Journal of Archaeological Science · March 2012

DOI: 10.1016/j.jas.2011.11.013

CITATIONS READS

4 206

2 authors, including:

Moujan Matin
Wolfson College, University of Oxford
8 PUBLICATIONS 18 CITATIONS

SEE PROFILE

All content following this page was uploaded by Moujan Matin on 09 May 2016.

The user has requested enhancement of the downloaded file.

 Journal of Archaeological Science 39 (2012) 763e776

Contents lists available at SciVerse ScienceDirect

Journal of Archaeological Science

journal homepage: http://www.elsevier.com/locate/jas

Egyptian faience glazing by the cementation method part 1: an investigation of

the glazing powder composition and glazing mechanism
Mehran Matin, Moujan Matin*
Research Laboratory of Shex Porcelain Co., 19th St., Kaveh Ind. Township, Saveh 39141-76314, Iran

a r t i c l e i n f o a b s t r a c t

Article history: In cementation glazing, for various reasons, there are significant differences between the composition of
Received 10 May 2011 the glass phases present in the faience microstructure and the glazing powder composition. Further-
Received in revised form more, in this glazing method, the glazing powder should perform more functions, which are: producing
19 November 2011
a shiny and smooth glaze coating, producing a firm capsule surrounding objects entirely (to prevent
Accepted 19 November 2011
adhesion between sintered glazing mixture and the melting glaze coating), and, finally, producing
a porous and friable sintered glazing mixture for safe and easy removal of buried objects. It is obvious
Keywords:
that these functions also have significant effects on the glazing powder composition and constitution.
Egyptian faience
Cementation
Therefore, analytical data on faience glass phases are very unlikely to reflect the strict composition and
Qom technique constitution of the original glazing powder properly. Accordingly, the composition and constitution of
Glazing mechanism cementation glazing powder have remained obscure for a long time. These data are particularly crucial
Glazing powder factors for cementation replication experiments and, consequently, for our awareness of the ancient
Glazing mixture faience production. In this paper, a series of cementation replication experiments have been conducted in
Capsule an attempt to provide further clarification of these issues. In addition, as a comparative indicator for
alkali and copper vaporisation, a small piece of rock quartz was placed separately above the glazing
powder. After firing, the appearance of the faience objects and quartz pieces and the microstructures and
chemical compositions of a select group of samples, obtained using Scanning Electron Microscopy e
Energy Dispersive Spectroscopy (SEM-EDS), were investigated. The discussion has focused mainly on
issues such as the raw materials and their contribution in the glazing process, the criteria for determining
the faience glazing methods, the necessary quantity of alkalis for a successful cementation glazing, and,
particularly, the glazing mechanism.
! 2011 Elsevier Ltd. All rights reserved.

1. Introduction alkaline-lime-silica glaze, often coloured by copper, that provides it

with its characteristic bright blue-green hue.
‘Faience’ was the French term originally given to tin-glazed Faience was first produced in the late fifth millennium BC in the
majolica type earthenwares made at Faenza, in northern Italy, Near East and then in Egypt. Subsequently, its production spread to
during the 16th and 17th centuries. Subsequently, the meaning has Rhodes, Crete, Cyprus, Russia, Italy and central Europe, France and
changed both in the French and English languages. In the 19th Britain. In the Indus Valley, faience was only produced during
century, European archaeologists used the misleading name of a short period, from the third millennium to the early second
‘Egyptian faience’ to designate ancient siliceous ceramics discov- millennium BC. Its production in Egypt continued until the Roman
ered in Egypt. period. Regarding the Near East, it is generally accepted that faience
Egyptian faience, described as “the first high-tech ceramic” was being produced there until the end of the Sassanian period
(Vandiver and Kingery, 1987), is basically a siliceous vitrified and (637 AD) (Tite and Shortland, 2008: 17, 93). However, it seems that
glost ceramic. Its non-clay body is composed mainly of finely faience production has never ceased in Iran (Persia). There are
ground quartz (rock or pebble) or sand, which is coated with an plenty of literature documents in Persian mentioning the produc-
tion of faience (Dehkhoda, 1956e1978), for instance, in the poems
of Nassir Khusrau, Sanai, and Mowlana (Rumi) (11th, 12th and 13th
centuries AD, respectively). In the early decades of the last century,
faience was made at least in two sites located at Qom and Shushtar,
* Corresponding author. Tel.: þ98 21 88284861; fax: þ98 21 88288361.
E-mail address: moujan.matin@gmail.com (M. Matin). cities in the centre and southwest of Iran (Wulff, 1966). Presently

0305-4403/$ e see front matter ! 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2011.11.013
764 M. Matin, M. Matin / Journal of Archaeological Science 39 (2012) 763e776

the ancient production process is still employed at Qom to some

extent.
It is now generally accepted that faience can be glazed using
three different techniques, namely application, efflorescence, and
cementation (Vandiver, 1983). It should be noted that a combina-
tion of these methods might have been used for one object.
Application glazing has been the most common glazing method in
ceramic technology. In summary, it involves the application of
a glazing suspension consisting of water and ground glazing
ingredients, such as pre-fired glass and/or quartz, colorant
(frequently copper), alkali flux, and lime. The glaze slurry can be
applied to the faience body by painting, pouring, or dipping.
Application glazing was common for producing large objects such
as tiles, and it is accepted that it may be recognised by the probable
presence of brush marks, drips, and glaze running lines, as well as
by considerable variations in the glaze thickness (Tite and
Shortland, 2008: 48e49, 208). The efflorescence method is a self-
glazing technique in which soluble alkali salts (carbonates, chlo-
Fig. 2. Faience object glazed by the cementation method.
rides, sulphates, etc.), copper compounds, ground quartz and water
are mixed to form a paste that is often shaped by moulding. During
the drying process, as the water evaporates, the soluble salts As pointed out above, visual examinations and related macro-
migrate to the surface of the object, and an efflorescence layer scopic evidence are the first approach employed for recognising the
(scum) develops. In the course of firing, this layer reacts with different glazing methods. However, these criteria may not be used
quartz, lime, and copper to form the glaze on the surface of the in many cases, for example when the object is a small fragment or
object. This method was used for the production of both small and the glaze is weathered. In order to provide more clarity in dis-
large artefacts. Objects made by this technique may show stand tinguishing between different glazing methods, Tite et al. (1983)
marks on parts which were set down during the drying process. and Tite and Bimson (1986) have argued that there are character-
Once again, variations in the glaze thickness are visible; the glaze is istic relations between the glazing method used and the micro-
thickest at the rims and on the outer surface, where the air flow and structure and composition of the fired object. Consequently, they
drying were the most significant. The cementation method is also proposed microstructural and analytical criteria for determining
a self-glazing technique. This method has been known to archae- which of the glazing techniques were used for the production of
ologists since the 1960s, when its use was first recorded at Qom, an a specific artefact. On the basis of these results, it was suggested
old town in Iran (Wulff et al., 1968); as such, it is sometimes that the macroscopic evidence should be supplemented by
referred to as the ‘Qom technique’. In this method, after the object microstructural and analytical evidence obtained from analytical
has been formed and dried, it is buried in a glazing mixture (the scanning electron microscopy (SEM). The faience microstructure
cement) consisting of alkalis, copper compounds, calcium oxide or consists of three different layers, which should be distinguished in
hydroxide, and/or quartz. Moreover, it is generally accepted that the object. The first, external layer is the quartz-free glaze layer. The
charcoal is one of the ingredients of the glazing mixture. During the second one is the interaction layer or ‘buffer layer’, as it is called in
firing process (at about 1000 " C), a glaze layer is formed on the ceramic technology (see e.g., Hamer and Hamer, 2004). This is the
faience surface as a result of different reactions between the glazing glazeebody interface, consisting of quartz grains surrounded by
mixture and the siliceous surface of the object. In this method, the glass matrix. The third layer is the body itself, containing varying
alkalis and copper vaporisation are crucial to the glaze formation. amounts of glass phase holding together the quartz grains of the
After firing, the partially sintered cementation mixture can be body. On the basis of laboratory replication (Tite et al., 1983; Tite
crumbled away from the object, which is now coated with glaze and Bimson, 1986), it is now widely accepted that objects made
(Figs. 1 and 2). Cementation glazing was appropriate for small by cementation glazing tend to have thinner glaze and buffer layers
objects such as beads. It is generally accepted that this method can and less interparticle glass. In contrast, efflorescence glazing has
be recognised by uniform glass thickness (Nicholson, 1993: 13; Tite a thicker glaze and buffer layers, and the greatest amount of
and Shortland, 2008: 48e49) and the absence of drying and firing interparticle glass. In the case of the application method, there is
marks for small objects. However, on larger pieces a rough glaze again less interparticle glass and a thin buffer layer, but the glaze
coating may be visible in the parts where they rested during firing. layer tends to be much thicker. Furthermore, in recent laboratory
replications, Tite et al. (2007) have investigated the composition
profiles of glass phases, from the glaze layer through the buffer
layer to the interparticle glass in the body, for objects produced by
the efflorescence and cementation glazing methods. The investi-
gated composition profiles for glazes formed by the efflorescence
technique mostly indicated an increase in the soda content and
corresponding decrease in the potash, calcium, magnesia and
copper oxide contents from the interparticle glass (in the body)
through to the glaze layer, which reflects the preferential efflores-
cence of sodium carbonate in comparison with potassium, calcium
and magnesium carbonates, and copper oxide. In the case of
cementation glazing, a decrease in soda and copper content from
the glaze layer through the buffer layer has been observed. In
addition, the absolute percentage of potash was significantly higher
Fig. 1. Schematic diagram illustrating cementation glazing (after firing). in both layers than in the original glazing powder (Tite et al., 2007;
 M. Matin, M. Matin / Journal of Archaeological Science 39 (2012) 763e776 765

Tite and Shortland, 2008: 51e54). The significant difference In part 1 of our study (the present paper), the replication
observed in the composition profiles for copper oxide makes it experiments have been conducted mainly by using synthetic
a useful criterion for distinguishing between the efflorescence and chemicals in order to provide the background information and
cementation glazing methods. In this context, preferential vapor- a basis for the second paper (part 2). The replication experiments
isation of potash during the firing of wood ashes has been reported will again be conducted in part 2, but using predominantly natural
(Misra et al., 1993), which may be expected in cementation glazing. raw materials in order to obtain a more realistic view of ancient
Furthermore, Paynter (2001) has suggested that potash in contact faience production. In this respect, part 2 will mainly focus on
with silica is a more effective flux than soda. Therefore, in a probable alkaline flux source that had been widely available to
cementation glazing, it will be preferentially taken up into the ancient faience makers.
melting glass.
In cementation glazing, the increased absorption of potash and 2. Experimental procedure
the reduced intake of calcium and magnesia (in comparison with
soda) reflect the fact that the composition of glass phases in faience Replication experiments by the cementation method were
differs significantly from the glazing powder composition from undertaken with nine glazing powder recipes. After firing, the
which they were produced. In addition, there are further known resulting objects and partially sintered glazing powders were
factors at work; plant ashes as a source of alkali contain not only observed by the naked eye and/or with a low power binocular
alkali carbonates, lime, and magnesia, but also considerable microscope, and macroscopic evidence were reported. Further-
amounts of chlorides and sulphates. In the context of ancient glass more, investigations on some selected objects were carried out
compositions, Rehren (2008) has argued that during the melting using SEM images (BSE mode) and SEM-EDS analyses.
process, the majority of chlorides and sulphates (present in plant
ash) form a separate salt melt, and it is only carbonates, bicar-
bonates, hydroxides, sulphites, and sulphates that are readily 2.1. Raw materials
incorporated into the glass melt. He states that “soda-lime-silica
glass and common chloride or sulphate-base salts form at melting Sand was used as a silica source for both faience bodies and
temperature a system of two immiscible melts, like oil and water” glazing powders. The particle size distribution and chemical
(Rehren, 2001: 484). Furthermore, laboratory replication experi- compositions of the raw materials are provided in Tables 1 and 2,
ments have proved that in comparison with soda, potash is respectively.
enriched in the salt melt and depleted in the glass melt, irrespective
of whether the chlorides were originally present as sodium or 2.2. Faience bodies
potassium chloride (Rehren, 2008; Tanimoto and Rehren, 2008;
Tite and Shortland, 2008). Recipes of faience bodies and their calculated chemical
As pointed out by Tite et al. (2007), cementation glazing pres- compositions are provided in Tables 3 and 4. Fine silica was used in
ents the advantage that a large number of small objects can be all samples except B2, in which coarse silica was used. As a binder
glazed and produced economically, and is thus appropriate for the and plasticiser, 5% Serish was used in all the recipes. Serish is the
mass production of small objects. It should be considered that in Persian name for the powdered root tuber of Asphodelus sphaer-
the mass production of faience objects by the cementation method, ocarpus, whose aqueous solution was commonly used as an adhe-
in actual practice, three significant features should be maintained sive in many traditional crafts in Persia, such as bookbinding and
by the glazing powder: first, the object should be covered by a shiny textile crafts (Wulff, 1966). In addition, its usage in the faience body
smooth glaze. Second, no adhesion should occur between the mixtures of the last century has been pointed out by Yaghobi
faience glaze coating and the glazing powder. Finally, the sintering (2002). The ash content of Serish is only about 5% at 900 " C. For
of the glazing powder should be optimum in order to provide safe better homogenisation, the resulting powder mixtures, which
and easy removal of the objects. Obviously, the mentioned features consisted of sand, Serish, etc., were sieved twice (30 mesh, 590
are affected by different factors in the production process, such as microns). By adding about 30 wt% water to each mixture, a paste
the ingredients’ particle size, the firing schedule, and, particularly, was created, which was wedged and kneaded for a few minutes
the glazing powder composition. All these factors should be care- without any further ‘aging’ of the body paste. The pastes were then
fully considered in the laboratory replication experiments; other- formed in two plastic (epoxy resin) moulds. The consequent shapes
wise, any results based on replication experiments may lead to were tabular discs of 40 mm in diameter that were made in two
misinterpretations. Therefore the attainment of these three thicknesses, 7.5 mm and 12 mm, for each body sample. The disc-
features should be regarded as a criterion for a successful cemen- shaped faience bodies were then left to dry at room temperature
tation glazing replication. for 48 h. The analytical data on the composition of the quartz
In summary, the composition and constitution of glazing pebbles and sands that had been used in ancient faience production
powder in cementation method have remained obscure for a long describes an alumina content of up to about 3.5 wt% (e.g., Tite and
time. Because of the reasons mentioned above, the analytical data Shortland, 2008: 38). Wulff et al. (1968: 100) have reported 5 wt%
on faience glass phases are very unlikely to reflect the exact Al2O3 and 1% Na2O in the faience body (before firing). Hence, to
composition and constitution of the original glazing powder. About replicate the aforementioned contaminant, 10 wt% feldspar and
half a century ago, in their seminal study, Wulff et al. (1968: 107) 5 wt% kaolin were added to the BR and B3 recipes, respectively.
stated that “systematic composition of the glaze powder from
these materials [pure materials] may eventually show which of the
Table 1
components of the glazing powder are essential and which are only
Particle size distribution of raw materials (microns, wt%).
incidental”. Therefore, this paper aims to improve present aware-
ness about the cementation glazing process, its technology, and Raw materials Less than 44 44e88 88e149 149e250
other relevant issues. To this purpose, hundreds of cementation Silica sand (fine) 40 60 e e
glazing replication experiments (not published) have been con- Silica sand (coarse) 10 20 25 45
Feldspar 40 60 e e
ducted by the authors, of which a select group is discussed in Calcium carbonate 100 e e e
this paper.
766 M. Matin, M. Matin / Journal of Archaeological Science 39 (2012) 763e776

Table 2 Table 4
Chemical composition of raw materials (wt%). Calculated chemical composition of faience bodies (wt%).

Raw materials SiO2 Al2O3 Fe2O3 TiO2 CaO MgO Na2O K2O L.O.I Code SiO2 Al2O3 Fe2O3 TiO2 CaO MgO Na2O K2O
Silica sand 95.66 1.68 0.17 0.05 0.7 0.01 0.03 0.45 0.73 BR 94.5 3.06 0.17 0.07 0.78 0.01 0.45 0.96
Feldspar 71.69 15.1 0.25 0.34 1.4 0.01 4.1 5.01 1.9 B1 & B2 96.87 1.7 0.17 0.05 0.71 0.01 0.03 0.46
Calcium 1.54 0.11 0.11 0.01 54.3 0.75 0.2 0.03 42.8 B3 95 3.49 0.2 0.06 0.71 0.02 0.49 0.03
carbonate
Kaolin 47.1 36.7 0.85 0.17 0.36 0.25 0.03 1.02 12.9

glazed faiences when crumbled (Fig. 6). No adhesion of the glazing

powder to the objects was observed. The faience body made with
2.3. Glazing and firing the B2 composition, about 6.5 mm in thickness, was entirely
translucent (Fig. 7; relevant analytical data and SEM images are
The raw materials used in the glazing powders were silica sand shown in Table 8 and Fig. 9 and will be investigated later in Section
(fine and coarse), calcium carbonate and hydroxide, alkali carbon- 4); however, the thicker sample, about 11 mm in thickness, was not
ates, potassium nitrate, common salt, copper oxide, and charcoal. translucent. Glaze runs and drip marks around the thick objects
The particle size of sand and calcium carbonate and their compo- were extensively observed, particularly on the objects whose
sition are provided in Tables 1 and 2, respectively. The grain size of position was not exactly horizontal during firing. Similarly, the rock
charcoal was less than 0.25 mm. The calcium hydroxide was quartz on the prop was covered entirely by a blue glaze exhibiting
produced by calcination of calcium carbonate at 980 " C. The runs and drips (Fig. 8a). Colonies of needle-like crystals were
calcined lime was then moistened and left to dry for 15 days at observed on the thicker patches of quartz glaze coating (Fig. 8b).
room temperature.
The compositions of glazing powders are summarised in Table 5.
3.2. GP1
The ingredients of glazing mixtures were weighed and mixed, and
the resulting powders were sieved twice (30 mesh) and placed into
Faience objects glazed by GP1 were covered by a shiny coating.
unglazed porcelain containers with a stood prop in their centre
However, the colour they exhibited was mostly white, and only
(Fig. 3). Then the faience bodies were buried alongside each other
blue in patches (Fig. 5a). The quartz piece was entirely white and
in the middle height of the glazing powder and around the prop. A
unglazed. The glaze coating was even and smooth on the BR and B3
piece of rock quartz was placed on top of the prop as a relative
samples, but wavy and uneven on the B1 and B2 samples. The
indicator for alkalis and copper vaporisation, and, finally, every-
reverse sides of the objects, where they were rested during firing,
thing was covered by a container lid. An orifice of 25 mm in
are almost unglazed, particularly on the central parts.
diameter had been made in the centre of the container lid, to model
the kiln chimney (similar to actual mass production firing). All
samples were fired in such lidded containers; however, the GP5 3.3. GP2
sample was fired in containers both with and without a lid (to
ensure oxidation firing in the latter case). Firing was carried out in All faience bodies which were glazed by GP2 exhibited a rich but
an electric kiln, according to the firing schedule shown in Fig. 4 very uneven coating, including many holes penetrated to the
(980 " C maximum temperature). faience bodies. The glaze colour was not turquoise blue but
greenish navy blue (Fig. 5b). The quartz piece was entirely covered
by glaze coating of the same colour. The sintered glazing powder
3. Results was more porous than GPR and GP1. However, adhesion of the
glazing powder to the glaze coating, regardless of body types, was
3.1. GPR observed. The underside of objects, where they were rested during
firing, was covered by a rich glaze and there was no considerable
Based on the authors’ previous replication experiments (not difference between the underside and upside glaze coatings.
published), the GPR and BR samples have been our criterion and Runners and drips were extensively observed on the coating of the
reference for the glazing powder and faience body compositions, quartz piece. The appearances of the GPR, GP1 and GP2 products
respectively. As expected, the resulting faience was covered by are summarised in Table 6.
a rich shiny turquoise blue glaze (Fig. 5c). However, all disc-shaped
objects showed uneven and rough glaze coating on the central parts
3.4. GP3
of their underside, where they were rested during firing. The glaze
coating was smooth and perfect on bodies BR and B3, but wavy and
The observations were the same as for GPR, except for two
uneven on B1 and particularly B2. Furthermore, on the glaze coating
cases; firstly, the sintered glazing powder was too dense for an easy
of some objects (different bodies), small crystals were observed in
removal of the objects. Secondly, there were patches of glazing
the glaze matrix. Optimum sintering was exhibited by the glazing
powder stuck on the glaze coating.
powder; it crumbled easily when pressed by hand. A capsule or
cocoon was formed completely around the objects, which released
3.5. GP4, GP5

Table 3 Macroscopic evidence observed on the objects glazed by GP4 and

The recipes of faience bodies (wt%).
GP5 (fired in the containers without lid e oxidation firing) were the
Code BR B1 B2 B3 same as for GPR products. However, it seems that the sintered
Fine silica sand 85 95 e 90 glazing powder was slightly porous and friable. In contrast, the
Coarse silica sand e e 95 e objects produced by GP5 in the lidded container indicated an
Feldspar 10 e e e unsuccessful glazing process; the glazing powder was found
Kaolin e e e 5
Serish 5 5 5 5
adhering entirely to the glazed objects; in addition, almost no blue
colour was observed on the objects, and small red patches (known as
 M. Matin, M. Matin / Journal of Archaeological Science 39 (2012) 763e776 767

Table 5
Composition of glazing powders used for replication experiments.

Experiment Calculated composition of glazing powder base (wt%) Additivesb (g)

code
SiO2 {f} SiO2 {c} CaO {ch} CaO {cc} Na2O {s} K2O {pc} K2O {pn} CuO Al2O3 MgO Fe2O3 TiO2 NaCl Charcoal Na2O {s} K2O {pc}
GPR 29.15 e 59.31 e 3.71 2.44 e 3.34 0.62 1.33 0.07 0.03 4 e e e
GP1 29.15 e 59.31 e 3.71 2.44 e 3.34 0.62 1.33 0.07 0.03 e e e e
GP2 29.15 e 59.31 e 0.36a 0.22a e 3.34 0.62 1.33 0.07 0.03 10c e e e
GP3 29.15 e e 59.31 3.71 2.44 e 3.34 0.62 1.33 0.07 0.03 4 e e e
GP4 29.15 e 59.31 e 3.71 2.44 e 3.34 0.62 1.33 0.07 0.03 4 5 e e
GP5 29.15 e 59.31 e 3.71 2.44 e 3.34 0.62 1.33 0.07 0.03 4 10 e e
GP6 29.15 e 59.31 e 3.71 2.44 e 3.34 0.62 1.33 0.07 0.03 4 e 1.5 1.5
GP7 e 29.15 59.31 e 3.71 2.44 e 3.34 0.62 1.33 0.07 0.03 4 e e e
GP8 29.15 e 59.31 e 3.71 e 2.44 3.34 0.62 1.33 0.07 0.03 4 e e e

{f}: Supplied by fine silica sand as the main source.

{c}: Supplied by coarse silica sand as the main source.
{ch}: Supplied by calcium hydrate as the main source.
{cc}: Supplied by calcium carbonate as the main source.
{s}: Supplied by sodium carbonate as the main source.
{pc}: Supplied by potassium carbonate as the main source.
{pn}: Supplied by potassium nitrate as the main source.
a
Supplied by other ingredients and not by direct use of alkaline carbonates.
b
(g): added to 100 g of glazing powder base.
c
(g): added to 94.43 g of glazing powder.

‘copper red’ in ceramic technology) were noted instead, reflecting appearance of the faience had been investigated in the authors’
the reduction atmosphere during firing. Similarly, the quartz piece previous replication experiments (not published), and no marked
was glazed more or less by a white coating with small red patches. differences in appearance were registered. However, the colour of
faience glazes which were produced with a lower Na2O/K2O ratio
3.6. GP6, GP7, GP8 slightly tended to pale blue. Moreover, the GPR replication
demonstrated that in the glazing powder composition, only about
The objects glazed by GP6 exhibited lighter blue glazes (in 5e6 wt% of alkali oxides (supplied by carbonates) are required for
comparison with GPR products). In addition, the same colour was a successful cementation glazing. The practical importance of these
observed on the quartz piece coating. Other observations were the data lies in the fact that such a low amount of necessary alkalis
same as for GPR products. provides a wider range of options for selecting the sources of
The GP7 sintered glazing powder was more porous and friable. alkaline fluxes. Thus, these data open a new window into studying
Furthermore, patches of adhering powder were observed on the the possible alkali flux sources in ancient faience production. This
glazed objects regardless of body type. The other results were once issue will be discussed in more detail in part 2 of the present study.
again the same as for GPR products.
The evidence visible on GP8 products was the same as for GPR 4.2. Calcium carbonate or calcium hydroxide?
products and no considerable difference was observed between
them. ‘Lime’ is a general term that connotes calcium containing inor-
ganic materials in which calcium carbonate, oxide, and hydroxide
4. Discussion predominate. The broad use of this term in the scientific literature
on faience has been a source of much confusion in glazing tech-
4.1. Necessary quantity of alkali oxides in the glazing powder nology. In the GPR experiment, a successful cementation glazing
composition was conducted using calcium hydroxide. In contrast, in the GP3
experiment, calcium carbonate has been the source of CaO.
In the replication experiments mentioned above, the Na2O/K2O Consequently, as mentioned in Section 3, adhesion of glazing
ratio in the glazing powder composition was approximately 3 in all powder to the object coatings was observed. Furthermore, the
cases. The effect of the Na2O/K2O ratio, in a range of 0.1e10, on the sintered glazing powder was too dense for a safe and easy removal
of objects. These replication experiments demonstrate that for

Fig. 3. Setting of faience and quartz objects for firing. Fig. 4. Firing schedule for faience and quartz objects fired in an electric kiln.
768 M. Matin, M. Matin / Journal of Archaeological Science 39 (2012) 763e776

Fig. 5. Faience objects made with BR body paste and glazed by GP1 (5a), GP2 (5b), and
GPR (5c) glazing powders.
Fig. 7. Translucent replica of faience object (about 6.5 mm in thickness) made by GPR
glazing mixture and B2 body paste.
a successful cementation glazing, in actual mass production,
calcium carbonate cannot be used as a CaO source.
contribution of potassium nitrates in the cementation glazing
process. This issue will also be discussed in more detail in part 2 of
4.3. Charcoal
this study.
It seems that the use of charcoal as a raw material in cemen-
tation glazing has been first noted in the observations of Wulff et al. 4.5. Uniformity of the glaze thickness
(1968). In the GP5 experiments (fired in the lidded container), it
was observed that the reduction atmosphere has a destructive As pointed out in Section 1, the presence of a uniform glaze
influence on the glazing process, particularly in terms of furnishing thickness has been generally accepted as a characteristic of
the glaze coating with copper blue. In cementation glazing, copper cementation glazing. However, significant variations in glaze
vaporisation with contribution of chlorides is the determining thickness were observed on the objects produced by successful
factor in the production of copper blue hues on the glaze coating
(this topic will be discussed later, in Sections 4.7 and 4.8). The blue
Table 6
hue in glaze coating is caused by Cu2þ ions. In a reduction atmo-
Coating appearance on the objects made by GPR, GP1 and GP2 glazing powders.
sphere, according to the amount of available oxygen, reddish brown
hues caused by Cuþ or the bright red resulting from copper metal Body GPR GP1 GP2
(in heavy reduction conditions) can become visible, as has been Quartz - Smooth shiny - Unglazed - Smooth shiny
seen on the GP5 sample. turquoise blue glaze. deep greenish
- Colonies of blue glaze.
It should be taken into consideration that generally the ashes needle-like crystals - Colonies of
also contain residual carbon. The residual carbon in both ashes and observed. needle-like
charcoal may produce a reduction atmosphere during the firing - Runs and drip crystals
process, particularly in mass production. On the other hand, in the marks observed. observed.
- Runs and
GPR experiment, a successful cementation glazing was conducted
drip marks
without any use of charcoal. Thus, the question arises whether observed
charcoal was indeed one of the raw materials used in cementation extensively.
glazing. As observed in the GP4 and GP5 experiments, there is only BR - Smooth shiny - Shiny whitish - Very uneven
one advantage in using charcoal, i.e. a slightly more porous sintered turquoise blue glaze, pale blue and holey
glaze. only in patches. with shiny
glazing powder. It seems that this advantage cannot be
- Needle-like - Underside of deep greenish
a convincing argument for the use of charcoal in cementation crystals observed objects is almost blue glaze.
glazing. This issue will be discussed further in part 2 of this study. in the SEM image. unglazed or - No considerable
- Runs and drip roughly vitrified. difference
marks observed. between the
4.4. Potassium nitrates
- Underside of glaze coatings
objects more or on the upper
In the GP8 experiment, the potassium carbonate was replaced less glazed, but side and
by potassium nitrate, and, once again, a successful cementation with rough underside of
patches. objects. Both
glazing was observed, which practically demonstrates the potential
are glazed.
B1 - Uneven but shiny - Uneven whitish - The same as
turquoise blue glaze. glaze, pale blue above.
only in patches.
B2 - Uneven but shiny - The same as above. - The same as
turquoise blue glaze. above.
- Only thinner
objects were
translucent.
- Needle-like
crystals observed
in the SEM image.
B3 - Shiny turquoise - Shiny whitish - The same as
blue glaze. glaze, pale blue above.
- Runs and drip marks only in patches.
observed.
Fig. 6. Faience object surrounded by its capsule.
 M. Matin, M. Matin / Journal of Archaeological Science 39 (2012) 763e776 769

Table 7
EDS analyses of needle-like crystals (normalised to 100%).

Raw no. Crystal appearance Fig. no. Glazing Sample Crystal Crystal composition Glazing
mixture type position mechanisma
SiO2 Al2O3 CaO MgO Na2O K2O CuO Cl
1 Needle-like grey crystals 11g GPR Faience BR Buffer 87.6 1.1 0.9 0.1 1.5 6.3 1.3 1.2 Combined
2 Needle-like grey crystals 9b GPR Faience B2 Body 83.0 1.5 5.1 0.2 1.8 4.6 2.9 0.9 mechanism
3 Needle-like grey crystals 10a GP2 Quartz Glaze 89.9 0.6 1.0 bd 0.6 0.9 6.2 0.8
4 Needle-like white crystals 10b GP2 Quartz Glaze 80.9 0.6 1.5 0.1 2.8 2.5 9.9 1.7 CGMb

bd: below detection limits.

a
Discussed in Sections 4e8.
b
Chloride Glazing Mechanism.

replication experiments. For example, in the GPR experiment, all that in cementation glazing the main and essential factor in
samples exhibited glaze runs and drips (particularly on vertical furnishing the glaze with copper blue colour is the presence of
sides of the objects). Indeed, it was difficult to find even one sample alkali chlorides. More than forty years ago, Wulff et al. (1968: 106)
with fairly uniform glaze. These observations dispute the argument pointed out the relation between copper vaporisation and alkali
identifying uniform glaze thickness as a characteristic of cemen- chlorides. They stated that “We suspect that the 1% copper oxide in
tation glazing. the glazing powder reacts with the salt (NaCl) in the plant ash to
form cupric chloride (CuCl2) which decomposes to cuprous chlo-
4.6. Microstructural criteria ride (CuCl) and chlorine at 993 " C. We also assume that there is
a significant concentration of copper in the vapour phase at the
As pointed out in Section 1, a thin interaction or buffer layer and firing temperature in the kiln (1000 " C) to react with the newly
the absence of interparticle glass are widely accepted as criteria for formed glaze on the bead surface, turning the glaze colour into
cementation glazing. However, in the GPR experiment, using a deep turquoise blue. Only at spots where there is direct contact
coarser sand in the body mixture (B2), an entirely translucent with larger copper oxide particles is a direct colouring effected in
object approximately 6.5 mm in thickness was obtained, which is the form of the darker specks mentioned above.” However, in the
shown in Fig. 7 (in the authors’ previous unpublished replication authors’ previous replication experiments (not published), the
experiments, such translucent samples were observed in objects vaporisation of copper at a lower temperature (920 " C) was
that were fired with a glazing mixture containing 10 wt% alkali observed. In other words, if the GPR composition were fired at
oxide, at 920 " C). It is apparent that this observation reveals more 920 " C, the quartz piece would be covered by a blue glaze coating.
interparticle glass in the body of the B2 object. SEM images of the In the presence of alkaline chlorides, the excessive deposited
aforementioned sample show a glaze layer (GLZ) with variable copper, with contributions from alkalis, may crystallise as needle-
thickness, between 150 and 300 mm, and the absence of a buffer like crystals, as mentioned for the first time by Wulff et al.
layer (Fig. 9a), but the presence of a continuous glass matrix (light (1968). The presence of needle-like crystals in the continuous
grey) between quartz particles (darker grey) in the body (BDY, interparticle glass matrix in the B2 sample (glazed by GPR mixture)
Fig. 9b). Furthermore, needle-like crystals were observed in the was pointed out in the previous section (Fig. 9b). Further micro-
glass matrix, which will be discussed further in Section 4.7. This structural investigations on the objects produced by GPR and GP2
experiment indicates once again that, as previously mentioned by demonstrated the presence of these crystals in the glass matrixes.
Vandiver (1998), such microstructural criteria should be used with For instance, the needle-like crystals (darker grey) in the glaze layer
more caution. (GLZ) on the quartz object glazed by GP2 are shown in Fig. 10a. In
addition, at higher magnifications on the same image, the small
4.7. Alkali chlorides: the essential glazing powder ingredient (about1e3 mm) white needle-like crystals were detected in the
glaze (Fig. 10b). The analytical data obtained by SEM-EDS on typical
In the GP1 experiment, chloride elimination from the glazing crystal samples (both white and grey crystals) are shown in Table 7.
powder composition resulted in an unglazed and white quartz It should be noted that in the GP6 experiment, by using more
piece and an entirely white glaze coating with small blue patches alkali carbonates in the glazing powder composition, a pale blue
on the faience objects (Fig. 5a). These observations reflect the fact glaze coating was achieved both on the quartz piece and the faience

Table 8
EDS analyses of the glass phases of objects glazed by GP1, GP2 and GPR glazing mixtures (normalised to 100%).

Experiment Sample type Glass phase Raw no. Glass phase composition Glazing
code position mechanism
SiO2 Al2O3 CaO MgO Na2O K2O CuO Cl
GP1 Faience BR Glaze 1 61.8 1.7 0.6 bd 6.2 27.6 2.1 bd IGMa
Buffer 2 70.8 6.8 0.3 bd 3.8 17.6 0.7 bd
Body 3 50.9 18.9 0.1 0.2 4.6 24.8 0.5 bd
GP2 Faience BR Glaze 4 69.4 9.8 0.7 0.4 2.3 1.7 14.3 1.4 CGMb
Buffer 5 80.3 9.5 1.4 0.2 2.1 1.9 3.4 1.2
Body 6 85.1 7.8 0.5 0.2 1.2 2.0 2.8 0.4
Rock quartz Glaze 7 80.8 0.4 0.1 0.4 4.8 0.6 10.9 2.0
GPR Faience BR Glaze 8 67.2 2.0 0.6 bd 5.1 15.6 8.4 1.1 Combined
Buffer 9 72.0 7.3 2.4 0.3 0.7 15.7 0.9 0.7 mechanism
Body 10 75.7 8.7 1.6 bd 0.6 12.7 0.4 0.3
Faience B2 Glaze 11 68.6 0.7 4.2 0.1 4.9 13.6 6.5 1.4
Body 12 67.3 2.0 11.4 0.8 2.9 13.0 1.6 1.0

bd: below detection limits.

a
Interface Glazing Mechanism.
b
Chloride Glazing Mechanism.
770 M. Matin, M. Matin / Journal of Archaeological Science 39 (2012) 763e776

Fig. 8. (a): Rock quartz glazed by GPR mixture. (b): Macrophotograph exposing the
details of a crystal colony on the quartz glaze. The framed area in Fig. 8a shows the
magnified section.
Fig. 9. Microstructure of translucent sample B2, glazed by GPR mixture. (a): SEM-BSE
image of a cross-section through the glaze (GLZ) and body (BDY). (b): SEM-BSE image
showing extensive interparticle glass in the body. BDY: body; GLZ: glaze.
object, reflecting a decrease in copper vaporisation. On the other
hand, in the absence of alkali carbonates in the GP2 experiment,
a greenish navy blue was exhibited both by the faience objects and chlorides in the faience glazing process. This issue will be discussed
the quartz piece (Fig. 5b). Typically, these deep greenish blue tones further in the Section 4.8.
are associated with a high concentration of copper in the low alkalis
glass phases, as it can inferred from the analytical data presented in 4.8. The glazing mechanism in the cementation method
Table 8, rows 4e7. On the whole, these observations reveal
a converse linear relation between the copper vaporisation rate and To provide a clearer background for the present discussion, SEM
the alkali carbonates content. In other words, more alkali carbon- images of cross-sections of BR faience samples glazed by the GP1,
ates in the glazing powder result in less vaporised copper. This GP2 and GPR mixtures are shown in Fig. 11. In addition, EDS
parameter is one of the main factors that explain why the firing analytical data on glass phases present in the aforementioned
temperature must be about 1000 " C in cementation glazing; at samples are provided in Table 8.
lower temperatures, an increased amount of alkali carbonates is Macroscopic, microstructural and compositional evidence
required for glaze formation (e.g., at 920 " C, a concentration of reveal that in the cementation method, the glazing of faience
about 10 wt% alkali oxides may be needed), and therefore less objects is carried out by two different mechanisms, namely the
vaporisation of copper and consequently pale blue products. In Interface Glazing Mechanism (IGM) and the Chloride Glazing
summary, the observations derived from the GPR, GP1, GP2 and Mechanism (CGM).
GP6 experiments demonstrate that in cementation glazing, by
increasing the concentrations of alkali chlorides and carbonates in 4.8.1. Interface Glazing Mechanism (IGM)
the glazing powder composition, the vaporised copper content will This mechanism has for the most part been known and dis-
increase and decrease, respectively. cussed extensively by researchers. It involves the diffusion and
As mentioned in Section 1 and based on the recent investiga- migration of alkalis from the glazing powder to the siliceous faience
tions conducted by Rehren (2008) and Tanimoto and Rehren body, and subsequent interactions between them to produce glass
(2008), partial contributions of alkali chlorides to faience glazes is phases which cover the object surfaces as glaze coating, and then
expected. In the GP1 and GP2 experiments, both the glazed quartz penetrate into the body to produce the buffer layer and interpar-
pieces and glazed faience objects exhibited contributions of alkali ticle glass between the fine grains of quartz. It is obvious that the
 M. Matin, M. Matin / Journal of Archaeological Science 39 (2012) 763e776 771

between 50 and 100 mm (Fig. 11a). The buffer layer (BFR) has well
defined boundaries and is about the same thickness as the glaze
layer. The quartz particles (darker grey) are more or less uniformly
distributed in the body (pores appear as black). Fig. 11b shows the
microstructure of the body at higher magnification, in which the
presence of interparticle glass (IPG, light grey) between quartz
particles (Q, darker grey) is visible.

4.8.1.3. Glass phase composition. The glass phase compositions of

the glazes made by this mechanism are essentially ‘high alkalis-low
copper’. The typical glaze composition is provided in Table 8, row 1.
A comparison between the glaze, buffer layer and interparticle
glasses reveals a decrease in copper content from the glaze layer to
the body. This aspect supports the previous observations of
Vandiver (1998). The significant amount of potash in glass phases
confirms the strong preference of the siliceous body and glass
network for Kþ (rather than Naþ), as has been pointed out by Tite
and Shortland (2008: 51e53) and Paynter (2001).
It seems that needle-like crystals in the glass matrixes are not
present in the objects produced by this mechanism, or at least they
have not been observed in this investigation.

4.8.2. Chlorides glazing mechanism (CGM)

This mechanism is fundamentally a type of vapour glazing,
based on the vaporisation of alkali chlorides. It involves deposition
of glaze on the objects by vaporised copper and alkalis. The prod-
ucts of the GP2 glazing mixture (Table 5) were glazed by this
mechanism. This mechanism plays a leading role in furnishing the
faience objects with copper green-blue hues. The main character-
istics of this mechanism are as follows:

4.8.2.1. Object appearance. Extensively holey glaze coating

(regardless of the body recipe) presenting deep greenish navy blue
hues (Fig. 5b) is the characteristic appearance of faience objects made
by this mechanism. Furthermore, there is no considerable difference
between glaze coatings on the upper side of the object and its
Fig. 10. SEM-BSE image of a cross-section through a quartz object glazed by GP2 underside (where they were rested on the capsule). Indeed, it is
mixture. (a): The needle-like crystals (darker grey) in the glaze matrix (GLZ, light grey). generally difficult to identify the upper side and underside of faience
(b): The grey and white needle-like crystals at higher magnification. BDY: body; GLZ:
objects glazed by this mechanism. In addition, runs and drip marks
glaze.
are not discernible because of the holey glaze coating. On the cross-
section, a thick buffer layer and a very pale green-blue colour can be
occurrence of this mechanism depends on the direct contact observed with the naked eye throughout the unvitrified texture of
between the glazing powder and the surfaces of siliceous objects, the body. As compared to objects glazed by the IGM mechanism, the
and it operates exclusively in the interfaces, as its name indicates. mechanical strength of faience objects glazed by CGM is typically
The objects produced by the GP1 glazing mixtures e in the absence considerably lower, and, as a result, they break easily.
of chlorides e were glazed by this mechanism (Table 5). The major The rock quartz (standing above the glazing mixture during the
characteristics of this mechanism can be categorised under three firing process) exhibits a rich shiny smooth glaze with the same
headings, namely ‘object appearance’, ‘microstructure’ and ‘glass colour and extensive runs and drip marks, reflecting strong
phase composition’. vaporisation of alkalis and copper.

4.8.1.1. Object appearance. Glaze coatings on the objects glazed by 4.8.2.2. Microstructure. A widespread and thick buffer layer,
this mechanism are typically white or whitish blue; however, pale generally thicker than 1000 mm and up to 2000 mm, is a typical
blue/blue patches may also be observed (Fig. 5a). The coating is characteristic of the microstructure of faience objects glazed by this
normally shiny, without any evidence of glaze runs or drip marks, mechanism, as shown in Fig. 11c. However, neither the buffer nor
but may also be even and smooth or holey and uneven, depending on the glaze layers display a continuous uniformity in their thickness.
body paste recipes (Table 3). The underside of objects, where they The microstructure of the body (Fig. 11d) shows minimal inter-
were rested, generally exposes a semi-vitrified coating and appears particle glass. In addition, the distribution of quartz particles and
poorly glazed or unglazed (particularly in the central parts). The rock glass phase is markedly uneven and there appears to be no
quartz which is placed above the glazing mixture (without any direct uniformity in the body texture overall.
contact with the glazing mixture) exhibits an entirely unglazed In contrast to faience objects, there is no obvious buffer layer in
surface, which reflects the absence of any alkalis vaporisation. the microstructure of rock quartz, but an extensive glaze layer is
present, as shown in Fig. 10a.
4.8.1.2. Microstructure. The microstructure of the BR sample glazed
by the IGM mechanism is shown in Fig. 11a and b. The thickness of 4.8.2.3. Glass phase composition. In contrast to the objects glazed
the glaze (GLZ, light grey) is moderately even but thin, and varies by the IGM mechanism, the glass phases present in the glazes
772 M. Matin, M. Matin / Journal of Archaeological Science 39 (2012) 763e776

Fig. 11. SEM-BSE images of cross-sections through replicated BR samples. (a, b): Glazed by GP1 glazing mixture (IGM mechanism). (c, d): Glazed by GP2 glazing mixture (CGM
mechanism). (e, f, g): Glazed by GPR glazing mixture (combined mechanisms). BDY: body; BFR: buffer layer; GLZ: glaze; Q: quartz.
 M. Matin, M. Matin / Journal of Archaeological Science 39 (2012) 763e776 773

produced by the CGM mechanism are of the ‘low alkalis-high very important to note that in the objects glazed by this mecha-
copper’ type, as it can be inferred from the analytical data nism, the microstructure and the amount of interparticle glass
provided in Table 8, rows 4 and 7. heavily depends on the firing schedule and particle size of the raw
The glaze coating on the rock quartz objects extensively exhibits material (silica), as noted for instance on the B2 sample glazed by
needle-like crystals, as discussed above in Section 4.7 (Fig. 10a and GPR (Fig. 9).
b; Table 7, rows 3 and 4).
4.8.3.3. Glass phase composition. The composition of glaze coatings
4.8.3. Combined glazing mechanisms produced by combined mechanisms can be classified as belonging
It is obvious that in actual practice, the faience objects are nor- to the ‘high alkalis-high copper’ type. The typical composition of
mally glazed by neither of the aforementioned mechanisms, but by such glazes is provided in Table 8, rows 8 and 11. Generally, the
a combination of both. Accordingly, the features presented by these glass matrixes made by this mechanism may contain needle-like
mechanisms may differ from the cases mentioned above. Hence, it crystals, for instance in the interparticle glass (see Fig. 9b), buffer
appears useful to regard this case as an independent mechanism, glass (see Fig. 11g), or glaze (see Fig. 10a and b).
which is consequently referred to as ‘combined glazing mechanism’ For purposes of clarity, a comparison between the main char-
in this paper. acteristics of the different glazing mechanisms is provided in
Considering the GPR glazing powder ingredients (Table 5), the Table 9.
main characteristics of the relevant products glazed by ‘combined
glazing mechanisms’ are classified as follows: 4.9. Firing shrinkage of faience objects

4.8.3.1. Object appearance. Shiny glaze coating with turquoise blue In comparison with most ceramic objects, faience objects
shades is the typical appearance of these objects (Fig. 5c). The glaze produced by the cementation method show unusual firing
coating on the faience objects may be smooth and even or holey shrinkage behaviour. The main unusual aspects of their shrinkage
according to the body paste recipes (see Tables 3 and 4). The process are as follows:
underside of objects is broadly covered by glaze coating; however,
it may also be rough, especially in central parts. Runs and drip 1) There is no limitation to the rate of shrinkage; the amount of
marks are generally observed, particularly on the vertical sides of shrinkage for a determined glazing mixture and firing
the objects. temperature depends largely on the duration of the firing
In samples produced by this mechanism, the rock quartz is process, particularly the ‘soaking time’ (the term used in the
always glazed and exhibits the same colour as the faience objects, firing schedule to denote the maintenance of the maximum
as well as extensive runs and drip marks. Colonies of crystals may temperature for a period of time). As the firing time rises, the
be observed on the thicker part of the glaze coating (Fig. 8). shrinkage increases more and more, and no evidence of object
melting has yet been identified.
4.8.3.2. Microstructure. The microstructure of faience objects made 2) Considering the firing time and the object’s size, the rate of
by BR paste and glazed by combined glazing mechanisms is shown shrinkage may be remarkable. In exceptional cases, the
in Fig. 11eeg. Well defined glaze and buffer layers, both about shrinkage process may continue to such an extent that the object
200e450 mm in thickness, are displayed in Fig. 11e. Generally, there depletes entirely and the remainder is only an empty capsule.
is extensive interparticle glass (light grey) between quartz particles 3) The endless shrinkage process may occur at any temperatures
(darker grey), mostly distributed uniformly (Fig. 11f). However, it is between approximately 920 " C and 980 " C, and it should be

Table 9
The main characteristics of glazing mechanisms.

Specification Object Interface Glazing Mechanism (IGM) Chlorides Glazing Mechanism (CGM) Combined glazing mechanisms
Appearance Faience (BR) - Shiny, smooth, white or - Holey glaze coating with deep - Shiny, smooth turquoise blue glaze coating.
whitish blue glaze coating. greenish navy blue hues. - Runs and drip marks observed, particularly
- No evidence of runs or drips. - The recognition of runs and drips on the vertical sides.
- The underside of objects is is not possible due to extensive - The underside of objects is covered by
poorly glazed or unglazed. holey surface. glaze coating; however, it may be rough,
- Quite similar glaze coatings on particularly in central parts.
the upper side and underside of objects.
- A thick buffer layer on the cross-section,
visible with the naked eye.
- Low mechanical strength.
Rock quartz - Entirely unglazed. - Smooth, shiny, deep greenish navy - Smooth, shiny, turquoise blue glaze
blue glaze coating. coating.
- Extensive runs and drip marks. - Extensive runs and drip marks.
Glaze Faience (BR) - The glass phase of the glaze - The glass phase of glaze layer is - The glass phase of glaze layer is of the
composition layer is of the ‘high of the ‘low alkalis-high copper’ type. ‘high alkalis-high copper’ type.
alkalis-low copper’ type.
Microstructure Faience (BR) - Generally, thin glaze and - A thick but very uneven glaze and - Well defined glaze and buffer layers,
buffer layers with buffer layers, generally thicker with extensive needle-like crystals
obvious and than 1000 mm. on the latter.
defined boundaries. - Needle-like crystals in the buffer layer. - Extensive interparticle glass between
- Limited interparticle glass - Minimal interparticle glass, unevenly uniformly distributed quartz particles.
between uniformly distributed between irregular
distributed quartz textures of body.
particles.
Rock quartz - Entirely unglazed. - No obvious buffer layer, but a thick - No obvious buffer layer, but a thick
glaze layer containing needle-like crystals. glaze layer containing needle-like crystals.
774 M. Matin, M. Matin / Journal of Archaeological Science 39 (2012) 763e776

highlighted that in actual practice there is no evidence of object hand to the formation of a capsule around the object (the latter is
melting (such as body vitrification, melting deformations, discussed in Section 4.10). It is obvious that the presence of ‘inter-
observation of melted glasses, or adhesion of sintered objects faces’ between the siliceous body and the glazing powder is
to the glazing powder). a prerequisite for diffusion. Therefore, in the initial stages of the
4) The objects shrink extensively only in their thickness (or shrinkage process, silica migration occurs in all interfaces (Fig. 13a
height), but not in their length or width (see Fig. 12). The and b), whereas by proceeding with the process, the silica migra-
faience object shown in Fig. 12 has been made with BR body tion takes place only through the underside of the object (Fig. 13c
paste as a small brick-like shape (40 # 21 # 7.5 mm) and glazed and d). Thus, after the initial stage of the shrinkage process, the
by a GPR glazing mixture. It was fired according to the schedule objects shrink only in their heights or thicknesses.
shown in Fig. 4, but with an additional 20 min soaking time.

According to the mechanisms and main causes involved, the

firing shrinkage of faience objects can be categorised as follows:

4.9.1. Firing shrinkage caused by glass phases produced by the IGM

mechanism
The glass phase produced by the IGM mechanism has been
discussed in Section 4.8.1. This shrinkage mechanism has been
previously discussed by researchers. For instance, Kingery and
Vandiver (1986: 277) state that: “the sodium carbonate diffuses
to the quartz paste body, where it reacts, releasing carbon dioxide
and forming an alkali silicate.as the first liquid is formed, it
penetrates into the quartz powder [body] so that the bead shrinks
away from the surrounding powder to allow room for the shiny
coating to be formed.” As it was discussed in Section 4.8.1, it is clear
that direct contact between the glazing powder and the surfaces of
the faience object is a prerequisite for the production of glass phase
by this mechanism. However, it should be noted that after the
initial stages of the shrinkage process, there would not be any
contact between the glazing powder and surfaces of the faience
object (except for the base of the object, where it stands on the
glazing powder). Therefore, this mechanism is effective only during
the early stages of the shrinkage process.

4.9.2. Firing shrinkage caused by glass phases produced by the CGM

mechanism
In comparison with glasses produced by the ICM mechanism,
those produced by the CGM mechanism (Section 4.8.2) provide
small contributions to firing shrinkage, mainly due to the minimal
amount of interparticle glass (see Fig. 11c and d). Furthermore, the
resulting glass is of the low alkalis type (Table 8, rows 4e7), and,
thus, it is less effective as a flux of silica.

4.9.3. Firing shrinkage caused by depletion of the siliceous body

from the faience object
In the cementation method, diffusion and migration of silica
from the siliceous body of the faience object to the glazing powder
occurs during the firing process. This phenomenon leads on the one
hand to the endless shrinkage of faience objects, and on the other

Fig. 13. Schematic diagrams illustrating the depletion of a siliceous faience object,
caused by silica migration to and reaction with the glazing mixture. (a): Interactions
which cause the shrinkage of objects during firing. (b): The initial stage of the
shrinkage process, simultaneous with the formation of the capsule. (c): Silica migra-
Fig. 12. Broken capsule showing the approximate amount of a faience object’s tion through the object’s underside causes a decrease in the height of the object. (d):
shrinkage. By continuing the silica migration, the object shrinks more and more.
 M. Matin, M. Matin / Journal of Archaeological Science 39 (2012) 763e776 775

The data obtained by wet chemistry analytical methods, The BR and B3 samples exhibited a smooth and even glaze
provided in Table 10, support the observations and schematic coating. As shown in Table 3, they respectively contain feldspar and
diagram (Figs. 12 and 13) and demonstrate the silica migration to clay. According to the mineral compositions of feldspar and clay, it
and reaction with the glazing powder. The compositions of the can be concluded that in cementation glazing the presence of
glazing mixtures (after firing) are provided in rows 1 and 3 and alumina in faience body compositions in minor proportions (see
their associated capsules in rows 2 and 4, respectively. As shown in Table 4) improves the glaze appearance. Indeed, in the actual
Table 10, the silica contents in capsules are significantly higher than production of faience objects by cementation glazing, using too
those in the glazing mixtures. pure sand leads to uneven and wavy glazes, as noted on the B1 and
B2 samples in the GPR experiment.
The analytical data provided in Table 8 demonstrate the
4.10. The mechanism of capsule formation
contribution of aluminium ions in the glass networks (glaze, buffer
layer, and interparticle glass) associated with the BR sample
In the cementation method, a further aspect that needs to be
(Table 8, rows 8e10), whereas there is little alumina in glass phases
considered is capsule formation. During the firing process, the
associated with the B2 sample (rows 11 and 12). Aluminium ions in
capsule prevents adhesion of glazing powder to the surface of the
glass networks significantly increase their surface tension and
melting glaze coating. Thus, the capsule formation process plays
viscosity. On the other hand, the glasses which contain too low
a crucial role in a successful cementation glazing.
a concentration of Al3þ are very fluid. As a general rule, very fluid
Calcium oxide is a refractory material with a melting point of
glazes usually cannot produce even glaze coatings on the porous
approximately 2570 " C. In silicates however, it may act as a flux,
bodies, because they penetrate very easily and quickly into the
depending on its content in the mixture. In the replication exper-
pores.
iments mentioned above, the ratio of CaO/SiO2 has been more than
Overall, it appears that in cementation glazing, there are addi-
2, as shown in the calculated composition data in Table 5 or the
tional decisive factors contributing to glaze smoothness. For
analytical data in Table 10, rows 1 and 3. Generally, the melting
instance, in the GP2/BR experiment, despite the aluminium
points of cementation glazing mixtures with such ratios are too
contribution in the glass network (Table 8, rows 4e6) and conse-
high for producing dense sintered mixtures at about 1000 " C. Thus,
quently a more viscous glaze, the glaze coating was uneven and
in these cases, a friable glazing mixture will be achieved after firing
holey, as mentioned in Section 3 and Table 6. By drawing
that enables an easy removal of objects. However, by decreasing the
a comparison between the microstructures of the different samples
CaO/SiO2 ratio to a value lower than 1, a dense sintered mixture will
shown in Figs. 9 and 11, in conjunction with analytical data on glass
be obtained in actual practice. In other words, increasing the SiO2
phases (Table 8), it becomes apparent that in cementation glazing,
content in the glazing mixture leads to a significant reduction in the
the smoothness of the glaze coating is determined by complex
melting point of the glazing mixture and hence in the formation of
interactions between different decisive factors, such as the amount
a dense sintered material.
of glass phases present in the body, buffer and glaze layers, the
The aforementioned phenomenon is the main cause of capsule
nature and composition of the glass phases e which is in turn
formation. By diffusion of silica into the glazing mixture (and the
determined by body contaminations and particularly glazing
resulting reduction of the CaO/SiO2 ratio), a layer of dense sintered
mechanisms e, distribution and size of body particles, etc.
mixture forms around the object, which is known as a ‘capsule’.
As it is known in ceramic technology, the suspensions and
After the initial stage of the shrinkage process and the simulta-
pastes of non-plastic materials such as quartz or sand are dilatant.
neous formation of the capsule, silica migration continues only
When mechanically disturbed, the dilatant paste appears to stiffen,
through the object’s underside, at the bottom of the capsule. Thus,
but if the disturbing force is removed, it becomes mobile again.
the bottom or the base of the capsule is generally thicker than its
Dilatancy is a destructive factor in faience pastes, causing a reduc-
walls, as shown in the schematic diagram (Fig. 13c and d).
tion in their plasticity and ‘workability’ and limitations in the
As mentioned above, silica migration is the major cause of
forming process. By using aqueous solutions of plasticisers such as
capsule formation. However, as Table 10 shows, the capsule
Serish, gum Arabic and gum tragacanth, the plasticity (or more
composition is generally richer in alkalis, which leads to more
precisely termed ‘workability’) of paste will improve slightly.
reduction in the melting point of the mixture, and consequently
However, by mixing dry Serish powder with sand and then adding
facilitates the process of capsule formation.
the required water, the workability of the resulting paste improves
considerably, so that it may be wedged and kneaded (just like the
4.11. Body mixture clay-based pastes). Furthermore, these pastes can be easily formed
in moulds, and the mould patterns are reflected on the bodies in
The glaze coating on samples B1 and B2, as compared with BR more detail. It should be considered that, unlike the clay-based
and B3, was not even and smooth. The glaze on B2 was more holey pastes, the ‘aging’ of such pastes has a destructive effect on their
and uneven, which reflects the fact that by using finer sand in the workability, and thus it must be avoided in replication experi-
body, the faience glaze will be more even and smooth. ments; the rheological properties of the aqueous solutions (or

Table 10
Chemical composition of capsules and sintered glazing mixtures resulting from GPR and GP2 experiments, obtained by wet chemistry analytical method.

Experiment code Raw no. Sample Chemical composition (wt%) Glazing mechanism

SiO2 Al2O3 CaO MgO K2O Na2O CuO Cl Totala

GPR 1 Glazing mixture 26.2 0.6 60.7 1.1 1.6 3.1 3.4 1.9 98.6 Combined mechanism
2 Capsule 44.3 0.9 42.5 1.0 1.0 7.7 1.2 0.6 99.2
GP2 3 Glazing mixture 25.4 0.5 58.4 1.1 0.3 4.6 1.7 6.8 98.8 CGMb
4 Capsule 53.2 0.7 34.7 0.8 0.2 5.5 2.1 1.3 98.5
a
Fe2O3, TiO2, MnO: not analysed.
b
Chloride Glazing Mechanism.
776 M. Matin, M. Matin / Journal of Archaeological Science 39 (2012) 763e776

suspensions) of Serish change significantly in time. For instance, if reflected on this paste in more detail. The propensity of such
such a solution is allowed to stand for 2 or 3 h, its viscosity a paste for forming by a potter’s wheel, at least in the case small
decreases significantly, and the dilatancy tendency increases. More objects, should be studied in future research.
dilatancy means less plasticity and, consequently, lower 11) In the body mixture, alumina bearing contaminations, such as
workability. feldspar and clay minerals, improve the glaze appearance. It
Regarding the B3 paste, it should be mentioned that the pres- seems that in cementation glazing, using too pure silica in the
ence of 5 wt% kaolin in its recipe did not have any considerable body mixture leads to a low quality glaze coating.
effect on the paste workability.
Acknowledgements
5. Conclusions
This project was financed by Shex Porcelain Co. The authors
1) A concentration of only about 5 wt% of alkali oxides (supplied
wish to thank Mehdi Shokr-e-Khodaie for drawing our attention to
by carbonates) in the glazing powder composition is sufficient
the production of faience in early decades of the last century in
for a successful cementation glazing at 980 " C. In actual prac-
Shushtar, providing video data, and establishing a personal
tice, the requirement of only such a low amount of alkalis in the
communication with his grandmother, who was involved in
glazing mixture means that in ancient faience production,
producing faience. We would also like to thank Yazdan Ghassemi-
there could have been more sources for alkali fluxes having
nia for preparing raw materials and Mahmoud Nader for photo-
lower amounts of alkalis in their composition, and not only
graphs and illustrating the figures. We are also grateful to Prof.
soda rich plant ashes and natron, which both have significant
Thilo Rehren and two anonymous referees for their helpful
amounts of alkalis in their composition. In other words, a wider
comments, which have greatly improved the original manuscript.
range of options for selecting the sources of alkaline fluxes can
Special thanks go to Raul Carstocea for his invaluable help with the
be considered, and this aspect opens a new window into
final editing.
studying potential alkali flux sources. This issue will be inves-
tigated further in part 2 of the present study.
2) Calcium carbonate is not a suitable source for supplying CaO. It References
has a destructive effect on the glazing process. Calcium
Dehkhoda, A.A., 1956e1978/1325e1357. Loghat-nameh. Tehran.
hydroxide is a proper alternative. Hamer, F., Hamer, J., 2004. The Potter’s Dictionary, fifth ed. London.
3) There is no convincing reason for confirming the use of char- Kingery, W.D., Vandiver, P.B., 1986. Ceramic Masterpieces e Art, Structure and
coal as a raw material in the glazing powder. Technology New York.
Misra, M., Ragland, K., Baker, A., 1993. Wood ash composition as a function of
4) It appears that potassium nitrate, like the fully reactive alkali furnace temperature. Biomass and Bioenergy 4, 103e116.
carbonates, contributes to the cementation glazing process. Nicholson, P.T., 1993. Egyptian Faience and Glass. Shire Egyptology, Buck-
5) Our observations did not confirm the macroscopic criteria inghamshire (series18).
Paynter, S., 2001. The Development of Vitreous Materials in the Ancient Near East
employed for identifying cementation glazing. In contrast, and Egypt, Unpublished DPhil Thesis, University of Oxford, Oxford.
glaze runs and drip marks were extensively observed on the Rehren, Th, 2001. Aspects of the production of cobalt-blue glass in Egypt.
replicated cementation glazes. Archaeometry 43, 483e489.
Rehren, Th, 2008. A review of factors affecting the composition of early Egyptian
6) Translucent faience objects with 6.5 mm thickness at temper-
glasses and faience: alkali and alkali earth oxides. Journal of Archaeological
atures in the range 920e980 " C have been made by cementa- Science 35, 1345e1354.
tion glazing which, in this case, did not confirm the Tanimoto, S., Rehren, Th, 2008. Interaction between silicate and salt melts in LBA
microstructural criteria employed for the identification of glass making. Journal of Archaeological Science 35, 2566e2573.
Tite, M.S., Freestone, I.C., Bimson, M., 1983. Egyptian faience: an investigation of the
cementation glazing. Such observations again emphasise that methods of production. Archaeometry 25, 17e27.
microstructural criteria must be used with more caution. Tite, M.S., Bimson, M., 1986. Faience: an investigation of the microstructures asso-
7) Alkali chlorides are essential raw materials for the cementation ciated with the different methods of glazing. Archaeometry 28, 69e78.
Tite, M.S., Manti, P., Shortland, A.J., 2007. A technological study of ancient faience
glazing process. They contribute to the formation of glass from Egypt. Journal of Archaeological Science 34, 1568e1583.
phases (interparticle glass, buffer and glaze). In addition, they Tite, M.S., Shortland, A.J., 2008. Production Technology of Faience and Related
play a prominent role in furnishing the glaze coating with Vitreous Materials. Oxford University School of Archaeology: Monograph 72.
Oxford University School of Archaeology.
copper blue hues, because the vaporisation of copper basically Vandiver, P.B., 1983. Egyptian faience technology, Appendix A. In: Kaczmarczyk, A.,
depends on the presence of chlorides in the glazing powder. Hedges, R.E.M. (Eds.), Ancient Egyptian Faience: an Analytical Survey of Egyp-
8) The alkali carbonates limit copper vaporisation. Increasing the tian Faience from Predynastic to Roman Times. Aris and Phillips, Warminster,
pp. A1eA144.
concentration of alkali carbonates in the glazing mixture leads
Vandiver, P.B., 1998. A review and proposal of new criteria for production tech-
to a reduction in copper vaporisation. nologies of Egyptian faience. In: Colinart, S., Menu, M. (Eds.), La Couleur dans
9) Deposition of alkalis and copper vapours on the surfaces of l’Emaillage del’Egypte Ancienne. Edipuglia, Bari, pp. 121e139.
Vandiver, P.B., Kingery, W.D., 1987. Egyptian faience: the first high-tech ceramic. In:
faience objects is the major cause of runs and drip marks. Also,
Kingery, W.D. (Ed.), 1987. Ceramics and Civilization, vol. 3. American Ceramic
the aforementioned vapours play a leading role in glazing the Society, Columbus, Ohio, pp. 19e34.
object’s underside (where it stands on during firing). Wulff, H.E., 1966. The Traditional Crafts of Persia. Cambridge, M.I.T.
10) By adding about 5 wt% Serish powder to the body mixture, and Wulff, H.E., Wulff, H., Koch, L., 1968. Egyptian faience: a possible survival in Iran.
Archaeology 21, 98e107.
then adding water, a more plastic paste was obtained that was Yaghobi, M., 2002/1381. Gozareshi az sakhte mohrehaye sonati dar Qom. Name-I
easily formed and moulded. The mould patterns could be Qom 17 & 18, 145e149.

View publication stats