This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
 Author's personal copy

Journal of Archaeological Science 38 (2011) 2633e2643

Contents lists available at ScienceDirect

Journal of Archaeological Science

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

The Temple of Venus (Pompeii): a study of the pigments and painting techniques
Rebecca Piovesan a, *, Ruth Siddall b,1, Claudio Mazzoli a, 2, Luca Nodari c, 3
a
Dipartimento di Geoscienze, Universit di Padova, Via Gradenigo 6, I-35131 Padova, Italy
b
Department of Earth Sciences, University College London, Gower Street, London WC1E 6BT, UK
c
Dipartimento di Scienze Chimiche, Universit di Padova, Via Marzolo 1, I-35135 Padova, Italy

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

Article history: We report here on a study of 57 fragments of wall painting excavated from the Temple of Venus
Received 13 August 2010 (Pompeii). These samples were characterised by a wide range of analytical methods. Data showed that
Received in revised form the palette is varied, although not so broad as that found in other buildings in Pompeii, and is consistent
23 May 2011
with pigments used elsewhere in Pompeii and in the Roman Empire. It is composed of: natural earths,
Accepted 24 May 2011
minerals and rare articial pigments. Paintings are made up of thin paint layers (0.01e0.10 mm thick)
strongly adhering to the underlying preparation layer. Nonetheless, in some cases thicker layers (up to
Keywords:
0.40 mm) were recognised, often spread on other previous painting layers. Samples were also compared
Pompeii
Pigments
with the microstratigraphic criteria developed in Piovesan (2009) to distinguish wall painting tech-
Fresco niques. This comparison demonstrated that both fresco and lime painting techniques were adopted.
Lime-based painting techniques 2011 Elsevier Ltd. All rights reserved.
Wall painting

1. Introduction time of the eruption of Vesuvius in AD 79, it was still under reno-
vation (Curti, 2007, 2008).
Roman pigments, particularly from Rome and Pompeii, have The main aim of this work is the characterisation of the paint-
been studied using analytical techniques and a variety of pigments ings from the Temple of Venus with a multi-analytical approach
and mixtures have been recognised (Augusti, 1967; Fuchs and and to derive information about the artistic knowledge, technology,
Barat, 1997; Meggiolaro et al., 1997; Edwards et al., 2002; Walsh and probable trade routes of the Sannitic and the Roman in Pom-
et al., 2003). One of the most important questions in the study of peii. Within the frame of this research, we also had the opportunity
ancient paintings is the identication of pigments and paint recipes, to test a new Mssbauer portable spectrometer.
which give useful information about the material knowledge of
a culture and helps in identifying lines of trade. The site of Pompeii is
2. Sample preparation and analytical approach
the most studied of the Roman Period, but many areas and buildings
remain unexplored; the Temple of Venus is one of those. Venus is the
57 fragments of wall painting were collected predominantly
patron divinity of Pompeii and her temple is one of the most
from debris and oor lling material representing ve different
important buildings in the town, located in the south-west of the site
chronological units from Republican to Imperial times (Table 1). In
(Fig. 1). The present building itself underwent numerous phases of
two cases (samples LF057, LF117) samples were also collected from
reconstruction and renovation. The area was probably a place of
their primary location, i.e. from walls of the 3rd Century BC and the
worship since Archaic time and was occupied by the Sannites from
end of the 2nd Century BC respectively. They represent the main
the late 4the3rd centuries BC. Redesigned during denitive Roma-
colour palette used for the wall paintings of the Temple of Venus.
nisation in 130 BC and renovated during the Julio-Claudian Periods,
The dimensions of the fragments ranged from 1 to 10 cm2. Due to
the building was destroyed during the earthquake of AD 62. At the
the fragmentary nature and, in most of the cases, secondary loca-
tion of the samples, it was impossible to make any technological
comments about the wall paintings as a whole; nevertheless the
* Corresponding author. Tel.: 39 049 8279143; fax: 39 049 8279134. analysis of these samples is a useful method for the identication of
E-mail addresses: rebecca.piovesan@unipd.it (R. Piovesan), r.siddall@ucl.ac.uk pigments and recipes used in the Temple.
(R. Siddall), claudio.mazzoli@unipd.it (C. Mazzoli), luca.nodari@unipd.it (L. Nodari).
1
Tel.: 44 020 7679 2758.
The samples were observed under an optical microscope in
2
Tel.: 39 049 8279144; fax: 39 049 8279134. thick and thin section, and also as pigment dispersions. Thick
3
Tel.: 39 049 8275173. sections were polished and studied under the optical microscope

0305-4403/$ e see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2011.05.021
 Author's personal copy

2634 R. Piovesan et al. / Journal of Archaeological Science 38 (2011) 2633e2643

Fig. 1. Map of the archaeological site of Pompeii with the location of the Temple of Venus (Tempio di Venere).

using inclined reected-light optical microscopy (RL-OM) with an
adjustable optic bre system. This type of observation allowed the
determination of microstratigraphy and the main microtextural
features of the pigments (grain size, shape, roundness and degree
of dispersion of the fragments of pigment in the matrix). A small
quantity of each pigment was then powdered by scraping the
surface of the sample with a scalpel, and dispersed on a glass using
Cargille MeltMount (nd 1.662 at 25  C). Sections were then
analysed by optical polarising light microscopy (PLM), both in plane
(PPL) and cross-polarised light (XPL). Petrographic thin sections of
plasters were also prepared and used for the study of the white
painted layers and the preparation layers.
Secondary electron images were acquired using a Fei Quanta
200 environmental scanning electron microscope (ESEM) at the
C.S.G. Palladio s.r.l. of Vicenza (CSGP), with an EDAX Falcon Energy
Dispersive X-ray Spectrocopy (EDS) microanalysis detector. Images
were taken under low vacuum conditions, to avoid gold or graphite
coating of the samples. Semi-quantitative EDS microanalyses were
also performed to obtain information on the elemental composi-
tion of the pigment particles.

Table 1
Time distribution of the recipes used for the preparation of the paints. Boxes lled with a dotted colour refer to samples of uncertain age.
 Author's personal copy
R. Piovesan et al. / Journal of Archaeological Science 38 (2011) 2633e2643
Chemical composition of one of the yellow pigments was
determined with a Cameca SX 50 electron probe micro-analyser
(EPMA) at the Institute of Geosciences and Georesources, CNR,
Padua, with an acceleration voltage of 15 kV and a beam current of
10 nA, and using mineral phases as standards. In order to minimise
volatile loss and alkali (especially Na) migration, analyses were
performed using a defocused beam (30 microns in diameter),
moving the specimen during the analysis. The counting time on the
element peak and background was 10 s and 5 s, respectively.
The identication of the different crystalline phases present in
some of the pigments was carried out by X-ray powder diffraction
(XRPD) on a Philips XPert powder diffractometer at the Depart-
ment of Geosciences, University of Padua, with Ni-ltered Ka1,2
radiation of Cu from a graphite monochromator. Measurements
were performed at 40 kV and 40 mA.
Absorption spectra in the IR region (between 4000 and
400 cm1) were collected at the CSGP using a Thermo-Nicolet Avatar
370 Fourier transform infrared spectrophotometer with the aid of
a diamond cell for the powdered samples and a Thermo-Nicolet
Continuum micro-spectrometer for the study of thick sections (m-
ATR), both of these instruments working in Attenuated Total
Reection (ATR) mode. Infrared spectroscopy is useful both for the
determination of pigments and for the identication of organic
binders. When applicable, the m-ATR method was preferred to
powder analysis with the diamond cell. In samples with very thin
paint layers, the second method was the only way to obtain a useful
spectrum. In these cases, a small amount of pigment was collected
from the surface with a scalpel. Since the two approaches returned
the same results, we will indifferently refer to as FT-IR analyses
hereafter.
Three red and two yellow paints were also analysed by Mss-
bauer spectroscopy in order to investigate on the Fe-bearing phases.
A new non-destructive portable 57Fe Mssbauer spectrometer
working in air and using a g-ray backscattering geometry was tested
in this study. This instrument, projected and assembled in the
Mssbauer laboratory, at Department of Chemical Sciences, Univer-
sity of Padua (DCS), is made of a standard constant acceleration
spectrometer equipped with four proportional counters working in
backscattering geometry. The detectors allow the simultaneous
collection, rigorously under the same energy conditions, of four
spectra, which are then summed each other by a dedicated software.
Spectra were collected at room temperature for a relatively long time
(5 days or more) in order to improve spectrum statistics and
analyzed by the usual tting procedure.
In addition, to select the most suitable samples for the Mss-
bauer analysis, an X-ray uorescence (XRF) spectrometry (DCS) has
been performed on a Bruker ARTAX portable mXRF, equipped with
a Mo tube, able to analyse an area of 0.79 mm2 acquiring for 300 s
using a voltage of 20 kV and an intensity of 1500 mA, working
under a He ux.
3. Wall painting microstratigraphy and painting techniques
All the wall painting samples from the Temple of Venus were
initially examined from a microstratigraphic viewpoint. Petro-
graphic thin sections were analysed using PLM. ESEM-EDS analyses
were also performed to better dene the sequence of layers.
3.1. Supports
Painting supports were shown to vary in terms of the number of
layers present and the relative coarseness of the mortars used,
referred to herein as intonaco. According to Vitruvius, intonaco is
a set of nishing layers composed of lime and very ne-grained
2635
sand, overlying a series of thicker and rougher mortar layers. The
paint is applied to the topmost of the ne layers (Rowland, 1999).
The following three classes of intonaco were recognised and
classied based upon the petrographic composition of the aggregate:
3.1.1. Marmorino
Marmorino is the most abundant intonaco in the painting
samples (72%). Under petrographic analysis in thin section, mar-
morino displays a micrite-like matrix (Piovesan et al., 2009)
composed of crypto- and microcrystalline calcite, with an aggre-
gate:binder ratio of about 1:1. Aggregate consists of euhedral
crystals of calcite and dolomite, with rare crystals of diopside,
feldspar and clasts of scoria present. In some cases, the carbonate
fraction of the aggregate is composed of well-rounded fragments of
micritic limestone rather than sparry calcite. This type of intonaco
often appears composed of two or three layers, only distinguished
by the change of grain size across the different coats.
3.1.2. Cocciopesto
Few intonaco layers of cocciopesto were identied among the
analysed samples (19%). These show a speckled matrix (Piovesan
et al., 2009) composed of cryptocrystalline calcite and hydrated
calcium silico-aluminates, with an aggregate:binder ratio of 1:1.
The aggregate is mainly composed of angular fragments of ground
ceramic materials. Cocciopesto only occurs in red and yellow
painting layers from second half of the 2nd Century BC, 1st Century
BC and Julio-Claudian Period units. Volcanic inclusions in the
ceramics indicate a local derivation of this material (Grifa et al.,
2005; De Carolis et al., 2009).
3.1.3. Intonachino
This type of intonaco is rarely used as preparation layer of
a painting and it is present in only 6 samples (9%). Intonachino is
characterised by an almost pure crypto- to microcrystalline calcite
matrix, with sporadic lime lumps and an aggregate:binder ratio of
about 1:1. The ller is composed of fragments of volcanic scoriae and
volcanic rocks, frequently associated with crystals of diopside and,
less often, sanidine, plagioclase, biotite and melanite. In three cases
(LF057, LF027.R, LF027.G) this intonaco acted as the preparation layer
for yellow, red and grey painting respectively; while in other three
cases (LF014, LF101, LF108), all belonging to units from the second half
of the 2nd Century BC, it underlies a smoothing layer. The mineralogy
and petrography of the volcanic component are compatible with the
products of Somma-Vesuvius (Santacroce, 1987).
At the top of the microstratigraphic sequence one or more painting
layers have been recognised. Among the samples considered for the
present study, single-layer paintings are predominant; but some cases
of multi-layer paintings occur, with a maximum of three coloured
layers. The superimposition of different colours is generally connected
to the presence of drawings or decorations such as colour stripes or
gures. Usually, a homogeneous colour was applied as a preparation
layer, followed by the decoration, generally painted with a different and
often more precious colour. Most of the samples considered here, show
continuity between the top intonaco layer and the rst painting layer
when observed under BSE. On the contrary, all the second or third
painting layers, and rarely some single-painted layers (e.g. sample
LF035), display a clear discontinuity with the underlying layer. As
demonstrated in Piovesan (2009), this discontinuity, constituted by
a higher concentration of calcite and a lower porosity, is one of the most
powerful tools for the distinction of two of the most common lime-
based painting techniques: the fresco and the lime painting (e.g. mez-
zofresco). In paintings executed with the fresco technique, pigment is
applied as a suspension in water onto a fresh plaster, and xing occurs
during carbonation of lime, so that only one carbonation layer is formed
on the surface. On the other hand, in the lime painting technique,
 Author's personal copy
2636 R. Piovesan et al. / Journal of Archaeological Science 38 (2011) 2633e2643
pigment is mixed with a binding material such as limewash (lime and
water), and then applied on a dry lime plaster. In this case, a carbon-
ation layer is observed on the surface of the plaster, and a second one on
the top of the painted layer, derived from the limewash.
3.2. Painting techniques
According to the indications reported in Piovesan (2009) about the
distinction between fresco and lime painting techniques, a distinction
between the two techniques was observed in the samples from the
Temple of Venus. These distinctions are based on the microstrati-
graphic sequences, and the presence of carbonation layers. The
majority of the samples showed that the paint layer is well adhered to
the preparation layer, without any discontinuity, and that there is only
one carbonation layer at the top of the sequence of layers (Fig. 2a, c).
This suggests that paints were applied on a fresh surface and there-
fore the adopted painting technique was fresco (Piovesan, 2009).
In a few samples, instead, a thin carbonation layer between the
plasters and the painting lms is clearly present (Fig. 2b, d). This
evidence indicates that in these samples the pigments were spread
on a dry or partially dry substrate, probably mixed with limewash
(e.g. mezzofresco). In these cases two carbonation processes have
occurred, the rst on the surface of the plasters, the second on the
surface of the painting layers due to the limewash used as binder.
Nevertheless, the thickness of the samples and the dispersion of the
particles within the painted layer suggest that pigment was mixed
with limewash (Piovesan, 2009). Sometime, in the same samples
several painting layers occur. In these cases the rst layer, directly
in contact with the plaster underneath, are spread using the fresco
technique, whereas the following paint layers are made by adopt-
ing a lime painting technique.
Fig. 2. Reected-light photomicrographs of wall painting samples considered in this study. a) single-layer painting; b) multi-layer colour painting. BSE images of wall paintings
analysed in this study: c) fresco; d) lime painting. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)
4. Pigments and recipes
The colour palette used in the Temple of Venus was determined
by a combination of optical and electronic microscopy, FT-IR and
Mssbauer spectroscopy.
The identied pigments are described in the following text and
classied by colour. A list of the recipes (one or more pigments
mixed together to obtain a specic hue), the related mineral
composition and the names of the samples that belong to each
recipe, are reported in Table 2 and discussed in Section 4.7.
4.1. Black
Observation under RL-OM in thick section revealed homoge-
neous, opaque and thin black layers (in the range 0.02e0.07 mm)
made by indistinct ne-grained particles. In these samples
(LF022.N, LF026.N, LF033.N, LF125, LF005) two black pigments have
been recognised.
4.1.1. Carbon-based black e ame-carbon
The majority of the black paints are made of a very ne-grained
black pigment, which is opaque under PLM, and associated with
rare, colourless, aggregates of ne-grained particles with high
birefringence. This second component is calcite, which may either
be present to lighten the hue or be part of the binder. BSE images
did not allow distinguishing pigment particles, while EDS semi-
quantitative analyses performed on selected areas showed
a higher concentration of carbon (C) with respect to the underlying
preparation layers, the absence of phosphorous (P) e which is
present in bone black e and the absence of metal oxides (Fig. 3). FT-
IR analysis only revealed the presence of calcite, which is clearly the
 Author's personal copy

R. Piovesan et al. / Journal of Archaeological Science 38 (2011) 2633e2643 2637

Table 2
Preparation recipes.

Recipes No. of samples Identied pigments Samples

Major components Minor components

Black 1 4 Flame-carbon e LF022.N, LF026.N, LF033.N, LF125
Black 2 1 Burnt ochre e LF005
Black 3 1 Red ochre, ame-carbon e LF126.N
Red 1 13 Red ochre Yellow ochre LF013, LF030.R, LF033.R, LF034.R, LF038.R, LF040,
LF044.R, LF116, LF117, LF124, LF126.R, LF127.R,
LF128.R
Red 2 4 Red ochre, yellow ochre Clays, green earth, yellow glass LF023.R, LF025, LF027.R, LF106
Red 3 4 Cinnabar Red ochre LF022.R, LF024.R, LF031.R, LF032
Yellow 1 3 Yellow ochre Red ochre LF015, LF028, LF102
Yellow 2 9 Yellow ochre and red ochre Yellow glass, clays LF008, LF010, LF018, LF036.G, LF099, LF107, LF126.G,
LF127.G, LF128.G
Yellow 3 1 Red ochre, yellow glass e LF057
Yellow 4 1 Yellow and brown glass Yellow ochre LF007
Light blue 2 Egyptian blue Yellow ochre, yellow glass LF024.A, LF035.A
Green 1 4 Green earth Red and yellow ochre, Egyptian blue LF030.V, LF031.V, LF034.V, LF035.V
Green 2 1 Red and yellow ochre, Egyptian blue, green earth Brown glass LF036.V
Green 3 1 Green earth, Egyptian blue Red and yellow ochre LF128.V
Grey 3 Yellow, red and brown ochre, Egyptian blue Brown glass, cinnabar, green earth LF026.G, LF027.G, LF044.G
White 5 Lime, calcite e LF037, LF038.B, LF097, LF101, LF108

main constituent of the carbonate binder, and is characterised by 4.2. Red

the bands centred at 1400, 875 and 712 cm1 assignable to CO2 3 .
XRPD only identied calcite, indicating the absence of any other
crystalline phase in the black paints.
Microtextural features, composition and spectroscopic data are
consistent with the use of a pigment composed of amorphous
carbon, i.e. lampblack, a ame-carbon pigment. These pigments
show ne-grained particles formed by the combustion of coal,
wood, oil or other fuel able to produce soot (Eastaugh et al., 2004).
This pigment was mainly applied alone (recipe Black 1), but
small quantities of it were also recognised in paint layers of other
colour, such as some red and yellow ones, probably added to
produce darker hues.
4.1.2. Burnt ochre
Only in one sample (LF005) was a mineral pigment been rec-
ognised. Under PLM it displayed ne opaque grains characterised
by very weak, reddish-coloured halo in transmitted light, scattered
within calcite agglomerates. This indicates that the pigment has
natural origin, or that it was produced by heating natural red ochre
at low temperatures. XRPD conrmed the presence of low amounts
of iron oxide phases. BSE and EDS analysis showed rare Fe-rich
particles well distributed in the painted layer.
Fig. 3. EDS spectrum of a carbon-based black particle.
Red colours are those most commonly used in the Temple of
Venus (Table 2) and are represented by three recipes and two
different pigments. In most of these samples, RL-OM showed red
particles, sometimes sparse, sometimes abundant, unevenly
distributed in layers of thickness ranging from 0.01 to 0.30 mm.
4.2.1. Red ochre
The samples analysed in this work were all collected from
buried units and therefore they did not come into direct contact
with the pyroclastic ow of the AD 79 eruption, suggesting that the
observed colour is from original red ochre paintings, and not as in
other instances at Pompeii, goethite-rich paints which have
transformed to hematite-rich paints due to thermal alteration
(Eastaugh et al., 2004; Cioni et al., 2004).
In most of the samples, PLM study under plane-polarised light
identied translucent to opaque red, orange and brown particles of
various sizes, from very ne- to medium-grained, or aggregates of
high relief ne-grained crystals (Fig. 4). Under XPL the interference
colours are typically masked by the strong body colour. BSE images
revealed the presence of several mineral phases with high amounts
of Fe (20e50 wt%). These data suggest that the colouring agent is
red ochre. XRPD analysis conrmed this evidence identifying
hematite as the main mineral phase. FT-IR spectra showed the
characteristic bands of the carbonate at 1400, 875 and 712 cm1.
Most of the red samples are made of dark, red ochre, very rich in
hematite. A red powder discovered within the area of the temple
(LF040), displayed the same composition and high hematite
content. PLM, XRPD and FT-IR conrmed that this material was
dominated by hematite with a very low amount of calcite present,
suggesting that it represents an abandoned pigment deposit.
The samples analysed suggest that the majority of the red
colours were prepared using a natural ochre, unusually rich in
hematite, which was selected and then deliberately mixed by the
artists with other pigments, in order to obtain the desired hues. In
order to study hematite-bearing pigments, three samples of red
colour (LF013, LF126.R, LF127.R) were investigated by Mssbauer
spectroscopy. All the spectra present the typical sextets of oxide
species attributable to hematite (Fig. 5a) with parameters very
close to those of very pure and well crystallized hematite, as
reported by Murad and Johnston (1987). Only one of the analysed
 Author's personal copy
2638 R. Piovesan et al. / Journal of Archaeological Science 38 (2011) 2633e2643
Fig. 4. Polarising light photomicrograph of red ochre agglomerate (PPL). (For inter-
pretation of the references to colour in this gure legend, the reader is referred to the
web version of this article.)
Fig. 5. Backscattering Mssbauer spectra. a) sample LF127.R; dots: experimental data
points; solid line: hematite. b) sample LF015; dots: experimental data points; solid
line: microcrystalline goethite; dotted line: nanosized oxyhydroxide particles.
samples (LF013) showed also a paramagnetic doublet attributable
to silicate and/or nanosized oxide particles. Considering that g-ray
backscattering Mssbauer information derives from a layer as thin
as the painting lm and Si content measured on the surface by
non-destructive XRF is negligible, doublet is most probably due to
nanosized oxide particles.
4.2.2. Cinnabar
Four of the red samples (recipe Red 3 in Table 2) are mercury
sulphide, HgS (the mineral cinnabar). Under RL-OM theses
pigments appear as very thin, brilliant-red painted layers,
0.02e0.04 mm thick, with very ne-grained particles (Fig. 6). Under
PLM in PPL, particles exhibited strong red-orange body colour,
weakly pleochroic from pale to deep red-orange and with
extremely high relief. Particles sometimes display rectangular or
columnar habits but predominantly the shape is irregular with
conchoidal fractures. Under XPL particles glow a deep red due to
high interference colours masked by the body colour and the strong
internal reections. XRPD and FT-IR identied traces of cinnabar;
only small amounts of this pigment are necessary to obtain a bright
red painting. ESEM-EDS analysis showed several particles, which
were rich in Hg and S, conrming the presence of cinnabar, and the
presence of a carbonate binder.
In two samples (LF022.R, LF031.R) cinnabar occurred alone. In
the other two cases (samples LF024.R, LF032), cinnabar was used in
conjunction with hematite presumably to both extend this valuable
pigment and brighten hematite red.
4.3. Yellow
The majority of the yellow paintings appear under RL-OM as
a thin pale yellow layer from 0.02 to 0.05 mm thick, although in
some cases (samples LF107, LF0126.G) the painted layer is irregular,
with thickness up to 0.3 mm. Four recipes (Table 2) and two yellow
pigments constitute yellow colours.
4.3.1. Yellow ochre
PLM analysis under PPL showed a pigment made from trans-
lucent yellow particles with ne to medium grain size and
moderate relief, associated with few scattered medium-grained
orange particles. Under XPL, pigment grains display high birefrin-
gence, and interference colours are strongly masked by the body
colour. XRPD analysis identied calcite, goethite and other iron
hydroxides. FT-IR spectra showed the characteristic bands of
Fig. 6. Polarising light photomicrograph of a crystal of cinnabar (PPL).
 Author's personal copy
R. Piovesan et al. / Journal of Archaeological Science 38 (2011) 2633e2643
carbonate (1400, 875, 712 cm1). EDS revealed high concentration
of Fe (between 15 and 35 wt%), conrming that pigment is mainly
derived from iron oxide hydroxides.
Two samples (LF015, LF128.G) containing yellow ochre were
also studied by Mssbauer spectroscopy. Sample LF128.G presents
only a low eld (37.7 T) sextet typical of oxyhydroxide species. The
parameters well t those of goethite, as described by Murad and
Johnston (1987). In addition to the goethite sextet, sample LF015
shows a doublet (Fig. 5b) attributed to nanosized oxyhydroxide
particles because silicates were not detected on the painting
surface by XRPD, FT-IR, EDS and non-destructive XRF analyses.
4.3.2. Yellowebrown glass
In a few samples, glassy yellowebrown fragments, displaying
irregular shape and conchoidal fractures (Fig. 7), have been rec-
ognised both under PLM and RL-OM. Under XPL, these particles
were isotropic. XRPD and FT-IR did not recognised any crystalline
phase (except for the calcite of the binder), while EDS microanalysis
showed that pigment particles contained Si, Al, Fe, Mg, Ca, Na, K
and traces of other elements. EMPA microanalyses were performed
in situ on some glassy yellow grains and their compositions were
compatible with those of trachyandesitic-phonotephritic glasses
from Somma-Vesuvius (Fig. 8). This indicated that these glasses
used as pigments probably derived from a natural volcanic glass,
which was deliberately ground by the artist to produce a yellow
pigment. This pigment was not been previously recorded.
4.4. Blue
The paints bearing this pigment are pale blue, green and grey.
The former are characterised under RL-OM by the presence of ne-
to medium-grained blue particles (up to 0.07 mm) homogeneously
distributed in a thin layer with a thickness of 0.05e0.15 mm. The
last two by smaller blue particles (up to 0.02 mm in diameter)
mixed with medium-grained red, yellow, green, and brown parti-
cles in thick layers of 0.1e0.2 mm with a bluish-greenish shade.
4.4.1. Egyptian blue
PLM observations in PPL revealed the presence of blue particles,
pleochroic from blue to pale blue with high relief, occasionally
displaying reddish patches. In the majority of the samples, these
particles have anhedral habits and are angular; more rarely they
showed euhedral shape and visible cleavage. In XPL, the rare well-
crystallised blue fragments show high birefringence partially
Fig. 7. Polarising light photomicrograph of two particles of yellow glass (PPL). These
particles are isotropic in XPL. (For interpretation of the references to colour in this
gure legend, the reader is referred to the web version of this article.)
2639
masked by the body colour. When observed through a Chelsea
lter, all blue particles transmit a dull red colour. FT-IR spectra
showed, besides the typical carbonate bands, two medium-intense
bands at 1000 and 1056 cm1 and other two bands at 1162 and
1192 cm1 with low intensity, which are typical of the Egyptian
blue (Fig. 9).
Blue particles in BSE images were also distinguishable on the
basis of their atomic number contrast. Phases with high atomic
number, which appear bright in BSE images, were composed, in
order of decreasing abundance, of C, O, Si, Cu and Ca, which is
characteristic of the calcium copper tetrasilicate CaCuSi4O10 (anal-
ogous to mineral cuprorivaite), the main component of the Egyp-
tian blue (Fig. 10) mixed with the calcium carbonate (CaCO3) binder.
Slightly darker inclusions in BSE images display a similar compo-
sition but with higher silica content, suggesting the presence of
a glass (Fig. 10a). More rarely, and only in the pale blue samples,
a phase containing tin (Sn) has been identied in small inclusions
within the CaCuSi4O10 crystals, which was CaSnSiO5 (analogous to
mineral malayaite) (Fig. 10b). The identication of this component
in the Egyptian blue suggests that the source of the copper was
from bronze (Hatton et al., 2008). Egyptian blue particles in grey
samples showed tiny KeCaeFe-bearing inclusions (Fig. 10a) and the
absence of CaSnSiO5; these evidences suggest that the source of
Copper for the pigment preparation probably derived directly from
copper ore or from copper metal (Hatton et al., 2008), indicating
different manufacturing techniques for the production of Egyptian
blue. By the 1st Century AD, manufactories of Egyptian blue were
developed throughout the Roman Empire, however there is insuf-
cient archaeological evidence which may be reliably used to locate
manufacturing technologies and materials to industrial centres
(Tite et al., 1984, 1987; Eastaugh et al., 2004). Because of the
extremely ne grain size of the Egyptian blue particles in green
samples, more precise determination of their composition was not
possible.
4.5. Green
Only one green pigment was recognised, which is the main
colouring agent for all the green colour recipes (Table 2). The
majority of the green samples appear under RL-OM as a thin pale
green paint layer with few well-distinguishable particles and rare
yellow and red ne particles. The thickness of the painted layer is
constant; 0.03 mm. In one sample where the green colour is
applied on a previous red painting layer, thickness reaches about
0.05 mm.
4.5.1. Green earth
PLM study revealed abundant rounded and sub-rounded parti-
cles with a colour varying from bright emerald green to pale green
(Fig. 11a). Green particles display a low relief and are made of
a polycrystalline aggregate as also shown by mottled extinction
(Fig. 11b). In XPL they show moderate birefringence and second to
third order interference colours, sometimes masked by the high
body colours. All these characteristics indicate that the green
pigment is a green earth. XRPD identied calcite, aragonite, quartz,
and celadonite or glauconite the primary components of the green
earth. Celadonite and glauconite are green ferromagnesian silicate
mineral belonging to the mica group. Although associated with
very different geological environments, these phases are indistin-
guishable using PLM (Deer et al., 1992). The relative proportions of
the different chemical components in the EDS spectrum conrmed
the presence of at least one of these minerals. FT-IR spectra showed
the presence of carbonates, due to the CaCO3 of the binder (bands
centred at 1400, 875 and 712 cm1 assignable to CO3) and silicates
(bands centred at 1000 cm1 assignable to the SieOeSi stretching),
 Author's personal copy

2640 R. Piovesan et al. / Journal of Archaeological Science 38 (2011) 2633e2643

Fig. 8. Total Alkali v. Silica diagrams used in the classication of volcanic materials. The compositions of the yellow glass particles are indicated by circles and the range of
chemistries demonstrated by Vesuvian obsidians/glasses (Santacroce, 1987) is shown by the area shaded.

Fig. 9. FT-IR spectrum of Egyptian blue in pale blue layer.

 Author's personal copy
R. Piovesan et al. / Journal of Archaeological Science 38 (2011) 2633e2643
Fig. 10. Back-scattered electron (BSE) images of Egyptian blue particles: a) CaCuSi4O10 crystals with glass phase (grey inclusions) and KeCaeFe-bearing inclusions (black inclusions);
b) CaCuSi4O10 crystals with Sn-rich phase (lighter inclusion).
possibly related to the pigment. The sample did not provide
a usable FT-IR spectrum, peaks were not clearly dened, and
therefore identication of these phases cannot be rened beyond
the observations made using PLM. We conclude that the pigment is
mainly composed of celadonite or glauconite, or potentially
a mixture of both phases. Geological deposits of celadonite,
weathering from basaltic rocks and glauconite, from marine sedi-
ments, would have been potentially available to painters in the
region of Campania.
4.6. White
About twenty white fragments from the Temple of Venus have
been analysed, but only ve of those are composed by a white
paint; as opposed to marmorino plasters without any painting
layers. Only one white pigment has been identied.
4.6.1. Calcite and limewash
PLM petrographic analyses of the majority of samples iden-
tied abundant grains of calcite, rarely as crystalline sparite and
more frequently as cryptocrystalline micrite-like agglomerates
probably deriving from the carbonation of lime. The former
either display regular shape related to calcite cleavage, or form
anhedral, colourless crystals. When sufciently large, crystals
show highly variable relief and high birefringence. XRPD, FT-IR
and EDS also indicate that calcite is almost the only component
present.
Calcite was identied in all the white paints, and it has also
a wide distribution through the other colour paints, where it is
Fig. 11. Polarising light photomicrographs of green earth in PPL (a) and XPL (b). (For interpretation of the references to colour in this gure legend, the reader is referred to the web
version of this article.)
2641
present in various percentages as an extender. In thin section,
a thick white smoothing layer was applied on an underlying mar-
morino preparation layer (Fig. 12). When associated with another
coloured layer, the white paint always acts as a ground (Fig. 12).
4.7. Recipes
Paints were either composed of pure pigments or prepared as
admixtures of pigments following a series of sixteen recipes, to
obtain different hues and shades (Table 2). The majority of the
recipes are simple and involve one or two main pigments, some-
times with the addition of other minor components. In order to
make colours darker or lighter, especially yellow and red, carbon
black and limewash, respectively, have been used in admixture
with the appropriate pigment. In other cases, paints containing the
same pigments were admixed in different proportions according to
different recipes (i.e. recipes yellow 1 and red 1), obtaining a variety
of hues. Grey and green 2 are the most complex recipes,
including four different pigments and several minor components.
In the case examined here, grey was made by admixing yellow, red
and brown ochre, Egyptian blue, brown glass, cinnabar and green
earth in proportions to obtain the desired hue, all pigments
expected to be found on the artists palette.
An additional important remark is that some of the colours, such
as black, red, green and yellow, were prepared using different
pigments. The adoption of different recipes for the same colour may
suggest presence of several ateliers of painters working in the
Temple of Venus who had preferences for different methods for the
production of their colours. Chronological data (Table 1) show that
 Author's personal copy
2642 R. Piovesan et al. / Journal of Archaeological Science 38 (2011) 2633e2643
Fig. 12. Reected-light photomicrograph of a white smoothing layer on a marmorino
preparation with traces of red paint (sample LF038). (For interpretation of the references
to colour in this gure legend, the reader is referred to the web version of this article.)
the most common and inexpensive recipes also vary with time, i.e.
pure hematite in red 1, hematite and goethite in yellow 2.
Black 1 seems to be the most used recipe for the black paints at
least from the 1st Century BC to the 1st Century AD, while Egyptian
blue in pale blue paints seems to be exclusively used in the Imperial
Period, although this pigment was certainly used since the end of
the 4th Century BC as demonstrated by its identication as
a component in green admixtures.
5. Conclusions
The colour palette of the Temple of Venus is varied, although the
range of colours is not as wide as found in other buildings in
Pompeii (nine different pigments have been identied in this
study). The types of pigments recognised are consistent with those
observed elsewhere in Pompeii and throughout the Roman Empire,
and correspond with the palette described by Pliny and Vitruvius
(Rackham, 1999; Rowland, 1999). The volcanic yellowebrown glass
is an exception. Inexpensive and widely distributed pigments such
as yellow and red ochres, green earth and limewash, were identi-
ed with precious pigments such as Egyptian blue and cinnabar.
Mssbauer spectroscopy performed on both red and yellow ochres
conrmed that these pigments are basically composed of a single
phase (respectively hematite, goethite) and that Fe-bearing sili-
cates are missing. In two cases (samples LF013, LF015) the presence
of nanosized oxide and oxyhydroxide particles were also recog-
nised. It is also worth noting that Mssbauer spectra were acquired
using a newly designed g-rays backscattering Mssbauer portable
spectrometer, which was tested on wall paintings for the rst time.
Statistics of the spectra were sufciently good and acquisition time
reasonably short to conrm the applicability of this new facility.
For Egyptian blue, the identication of CaSnSiO5 in pale blue
paints and that of an unspecied KeCaeFe-bearing phase as tiny
inclusions in pigment grains of grey paints, indicates two different
manufacturing processes in the preparation of this pigment. The
occurrence of CaSnSiO5 indicates that Cu was derived from bronze,
while the absence of this phase probably suggests a different origin
for Cu, such as directly from ore deposits.
A yellowebrown pigment was recognised and characterised for
the rst time. Composed of a yellow or brown volcanic glass, it is
chemically compatible with volcanic products of Somma-Vesuvius.
This suggests a local production for this pigment which is
undocumented elsewhere. This is relatively common in many
paints either as one of the main components (yellow 3, yellow 4) or
as a minor component (red 2, yellow 2, pale blue, green 2).
The nine pigments were used either in a pure form or in
admixtures in different proportions of two or more compounds;
sixteen recipes have been recognised. It is worth noting the grey
recipe uses seven pigments. Such a complex recipe for the
production of a grey colour has not been previously recorded,
although presumably such mixtures were not uncommon.
For the preparation layers, three different types of intonaco
have been recognised, on the basis of the composition of their
aggregate (intonachino, cocciopesto, and marmorino). Although all
the intonaco layers are meticulously prepared, marmorino is real-
ised as a multi-layered plaster, and reects more closely Vitruviuss
recommendations.
All the data, and in particular the absence of a carbonation layer
between the paint and the intonaco and of any trace of organic
binders, conrmed that the most utilised painting technique was
fresco. All samples display very thin painting layers, with well
dispersed ne-grained pigment particles and sometimes medium
to coarse grains well-coated by a lm of calcite. In multi-layered
coloured samples, when stripes or gures were painted on a col-
oured background, thickness of the painted layers and the presence
of a carbonation layer between the two differently coloured layers,
suggests that the more supercial painted one was applied with
a lime painting technique.
In conclusion, the importance and prestige of the Temple of
Venus is also reected in the quality and value of the pigments, and
in the careful execution of the mortars.
Acknowledgements
Many thanks to the Soprintendenza archeologica di Pompei, and
to Prof. Emmanuele Curti, Director of the archaeological excavation
at the Temple of Venus, for providing the samples and also for his
helpful discussion of the archaeological contexts. The authors are
extremely grateful to Dr. Paolo Cornale and the C.S.G. Palladio s.r.l.
for sponsoring and supporting this research. We also thank Dr.
Celestino Grifa for sharing his knowledge of the eruption products
associated with Mt Vesuvius. Finally, but not least, we would also
like to thank the staff of the Dipartimento di Geoscienze (Padova)
and of the Istituto di Geoscienze e Georisorse, CNR (Padova) for
analytical and technical support. We would also like to thank two
anonymous reviewers for their useful comments on this text.
References
Augusti, S., 1967. I colori pompeiani. De Luca Editore, Rome.
Cioni, R., Gurioli, L., Lanza, R., Zanella, E., 2004. Temperatures of the AD 79 pyro-
clastic density current deposits (Vesuvius, Italy). Journal of Geophysical
Research 109. doi:10.1029/2002JB002251 B02207.
Curti, E., 2007. La Venere Fisica trionfante: un nuovo ciclo di iscrizioni dal santuario
di Venere a Pompei. In: Il lo e le perle. Studi per i 70 anni di Mario Torelli.
Venosa, pp. 57e71.
Curti, E., 2008. Il tempio di Venere Fisica e il porto di Pompei? In: Guzzo, P.G.,
Guidobaldi, M.P. (Eds.), Nuove ricerche archeologiche nellarea vesuviana (scavi
2003e2006), pp. 47e60. Roma.
De Carolis, E., Grifa, C., Langella, A., Morra, V., Soricelli, G., 2009. Ceramiche tardo
antiche da Pompei. In: Actes de la Table Ronde Les cramiques communes de
Marseille Gnes (et en Languedoc, Campanie, Sicile), IIe s. av. J.-C. - IIIe s. apr.
J.-C, (Structures de production, typologies et contextes indits), Naples, 2e3
November 2006.
Deer, W.A., Howie, R.A., Zussman, J., 1992. An Introduction to the Rock-forming
Minerals, second ed. Longman Scientic & Technical., Essex. 296e297.
Eastaugh, N., Walsh, V., Chaplin, T., Siddall, R., 2004. Pigment Compendium. A
Dictionary of Historical Pigments. Butterworth-Heinemann.
Edwards, H.G.M., de Oliveira, L.F.C., Middleton, P., Frost, R.L., 2002. Romano-British
wall-painting fragments: a spectroscopic analysis. The Analyst 127, 277e281.
Fuchs, M., Barat, H., 1997. Analyses phisico-chimiques et peintures murales
romaines Avenches, Bosingen, Dietikon et Vallon. In: Roman Wall Painting.
 Author's personal copy
R. Piovesan et al. / Journal of Archaeological Science 38 (2011) 2633e2643
Materials, Techniques, Analysis and Conservation Proceedings of the Interna-
tional Workshop on Roman Wall Painting, Fribourg, 7e9 March 1996. Institute
of Mineralogy and Petrography Fribourg University, pp. 181e191.
Grifa, C., Langella, A., Morra, V., Soricelli, G., 2005. Pantellerian Ware from Miseno
(Campi Flegrei, Napoli). Periodico di Mineralogia 74 (1), 69e86.
Hatton, G.D., Shortland, A.J., Tite, M.S., 2008. The production technology of Egyptian
blue and green frits from second millennium BC Egypt and Mesopotamia.
Journal of Archaeological Science 35, 1591e1604.
Meggiolaro, V., Molin, G., Pappalardo, U., Vergerio, P., 1997. Contribution to studies
on Roman wall painting materials and technique in Greece: Corinth, the
southeast building. In: Roman Wall Painting. Materials, Techniques, Analysis
and Conservation Proceedings of the International Workshop on Roman Wall
Painting, Fribourg, 7e9 March 1996. Institute of Mineralogy and Petrography
Fribourg University, pp. 105e117.
Murad, E., Johnston, J.H., 1987. Iron oxides and hydroxides. In: Long, G.J. (Ed.),
Mssbauer Spectroscopy Applied to Inorganic Chemistry, Vol. 2. Plenum Press,
New York.
Piovesan, R., 2009. Archaeometrical Investigations on Mortars and Paintings at
Pompeii and Experiments for the Determination of the Painting Technique.
Ph.D. Thesis, University of Padua, Italy.
Piovesan, R., Curti, E., Grifa, C., Maritan, L., Mazzoli, C., 2009. Ancient plaster tech-
nology: petrographic and microstratigraphic analysis of plaster-based building
2643
materials from the Temple of Venus, Pompeii. In: Quinn, P.S. (Ed.), Interpreting
Silent Artefacts: Petrographic Approaches to Archaeological Ceramics.
Archaeopress, Oxford, pp. 65e79.
Rackham, H., 1999. Loeb Classical Library (trans.). Harvard University Press, Cam-
bridge, Massachusetts. Pliny (1st cent AD/Rackham 1952). Pliny Natural History.
Rowland, I.D., 1999. Vitruvius the Ten Books on Architecture (trans.). Cambridge
University Press. Vitruvius (1st cent BC/Rowland 1999).
Santacroce, R. (Ed.), 1987. Somma-Vesuvius, Quaderni de La Ricerca Scientica
Progetto nanziato Geodinamica. Monograe nali, 8. CNR Roma.
Tite, M.S., Bimson, M., Cowell, M.R., 1984. Technological examination of Egyptian
blue. In: Archaeological Chemistry III Advances in Chemistry Series, Vol. 205.
American Chemical Society, Washington DC., pp. 215e242.
Tite, M.S., Bimson, M., Cowell, M.R., 1987. The technology of Egyptian blue. In:
Bimson, M., Freestone, I.C. (Eds.), Early Vitreous Materials, British Museum
Occassional Paper, 56. British Museum, London, pp. 39e46.
Walsh, V., Siddall, R., Eastaugh, N., Chaplin, T., 2003. Pigmenti di Pompeii: verso la
denizione di uno standard di riferimento per la ricerca sui pigmenti romani.
In: Scienze e archeologia Proceedings of Giornate di Studio: Le scienze e
lambiente e Le scienze chimico siche, Pompeii, 23 October and 19
November 2003. Universit del Sannio-dipartimento di studi geologici e
ambientali and Universit di Modena e Reggio Emilia-dipartimento di chimica,
pp. 181e196.