Journal of Cultural Heritage 1 (2000) 91 – 110

© 2000 Éditions scientifiques et médicales Elsevier SAS. All rights reserved

S1296-2074(00) 00164-3/FLA

Panselinos’ Byzantine wall paintings in the Protaton

Church, Mount Athos, Greece: a technical examination

Sister Daniiliaa, Sophia Sotiropouloua, Dimitrios Bikiarisa, Christos Salpistisb,

Georgios Karagiannisa, Yannis Chryssoulakisc*, Beth A. Priced, Janice H. Carlsone
a
‘ORMYLIA’ Art Diagnosis Centre, Sacred Convent of Annunciation, 63071 Ormylia, Chalkidiki, Greece
b
Aristotle University of Thessaloniki, Thessaloniki, Greece
c
National Technical University of Athens, Athens, Greece
d
Philadelphia Museum of Art, Philadelphia, PA, USA
e
Winterthur Museum, Winterthur, DE, USA
Received 15 March 1999; accepted 27 January 2000

Abstract – The sole surviving fresco paintings of Manuel Panselinos (13th century AD), one of the most celebrated Greek
iconographers of the Byzantine era, are located in the Protaton Church (10th century AD) on Mount Athos, Greece. This
paper presents an examination and technical analysis of 15 representative thematic scenes, covering an area of
approximately 65 m2, from these monumental works of art. The following exhaustive study and documentation of both
the original and the subsequently restored areas of the wall paintings were made possible by using various imaging
techniques, including visible and ultra-violet photography, infrared reflectography, colour measurements and representa-
tion. The chemical identification of the pigments, binders and layer stratigraphy was achieved through the use of visible
and ultra-violet fluorescence microscopy, micro-Raman spectroscopy, Fourier transform mspectroscopy (mFTIR), X-ray
diffraction (XRD), and scanning electron microscopy with energy dispersive analysis (SEM-EDS) and electron probe
microanalysis (EPMA). A collaborative analysis, its results demonstrate that the paintings were executed in both the true
fresco and lime-painting techniques. They have also established Panselinos’ choice of materials and colour palette. We
believe this study to be an important and necessary prerequisite for the future preservation and restoration of these unique
frescoes. © 2000 Éditions scientifiques et médicales Elsevier SAS

Keywords: Byzantine frescoes / palette / painting technique / non-destructive analysis / micro-analytical identification

1. Research aims the most important of the monumental wall paint-

ings on Mount Athos.
The frescoes of the Protaton Church, completed Paradoxically, the universally recognised value of
at the end of 13th century, are attributed to Manuel Panselinos’ paintings is in inverse proportion to the
Panselinos, one of the most celebrated Greek icono- limited scope of the related bibliography, which, in
graphers of the Byzantine era [1 – 4]. any case, is exclusively oriented towards their his-
Their thematic scenes, representative of the Mace- toric and aesthetic qualities.
donian School of Byzantine iconography, comprise a The main objective of this present experimental
unique ensemble and, at the same time, are among work is to reduce this lacuna, at least as far as the

* Correspondence and reprints: [email protected]

92 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110

study of the structure of the surface and underlying 2.2. Sampling analysis
colour layers is concerned.
Its content is focused on two main axes: The aforementioned data were then verified by i)
a) a description both of the structural com- mRaman and mFTIR spectroscopy, ii) elementary
ponents and materials used by the iconographer scanning electron microscopy analysis and electron
and of his technique, with a view to exposing probe microanalysis combined with optical mi-
the artistic value and to promoting the work in croscopy [13].
general; X-ray diffraction analysis was also used in order
b) an exposition of the paintings’ current physical to identify the crystalline structure of the lime mortar
state of preservation together with a detailed descrip- components.
tion of damage, chemical modifications and other
interventions, aiming at the effective restoration of
these frescoes, and of the monument as a whole, 3. Experimental section
which is in immediate danger.
An extensive study of certain representative ele- 3.1. Thematic scenes
ments in Panselinos’ style, and in the style of the
Macedonian School of iconography in general, has Fifteen entire thematic scenes were studied in
been undertaken using a combination of both non- complete detail. The seven chosen for this examina-
destructive analytical methodologies and sampling. tion were considered to describe most eloquently
All of the evidence gathered using non-destructive Panselinos’ painting technique.
methodology was verified and completed by that Panselinos developed his frescoes in four zones,
derived from sampling techniques. each 2 m high. The zones are numbered by re-
searchers vertically, from the ground (first zone) to
the ceiling (fourth zone).
2. Introduction The religious scenes depicted, following the Byzan-
tine iconographic tradition, have also been adjusted
2.1. Non-destructive analysis to the architecture of this edifice: a basilica without
dome.
In the visible area of the spectrum, photography A meticulous in situ observation of the walls was
and macrophotography were taken of certain the- undertaken. Neither vertical joinings nor limewash
matic scenes, together with reference colour panels. overlapping could be observed between adjacent
This procedure permitted control of the colorimetric thematic ensembles within any one zone. On the
fidelity of the colour slides taken and, when neces- contrary, horizontal coatings between successive
sary, correction of colours during digitisation, by zones, in vertical order, were evident and demon-
means of a scanner. strated that the painter’s work had followed a de-
Infrared reflectography was focused especially on scending progression, from the ceiling to the ground.
a) detecting the initial drawing, b) uncovering the The wall surface of each zone, therefore, was contin-
subsequent pentimenti, or later interventions, c) re- uously painted and the painter was able to produce
vealing the design finally adopted, and d) indicative an entire ensemble of several thematic scenes in as
pigment identification [5 – 9]. much time as the limewash remained wet.
Ultra-violet fluorescence photography in the visi- The following two criteria were used to select the
ble area of the spectrum permitted diagnosis of the thematic scenes for this study, in so far as the
surface condition, based on the different degrees of painting technique and the current state of the mon-
fluorescence of the materials. It also provided ument’s preservation as a whole are concerned: the
indications for the identification of certain pig- richness and the representative character of the data
ments, which exhibit fluorescence of a characteristic obtained.
hue [5]. Subjects under consideration:
Colour measurements were made for a) the study
of the chromatic effect of the dirt layer covering the – complex The Entrance of the Mother of
painted surface, b) the determination of the colour scenes: God (2nd zone)
palette, c) the study of the achieved colour harmony, The Resurrection (2nd zone)
and d) the registration of the characteristic reflec-
tance spectra of both the pigments and their mix- – simple St Artemius (1st zone)
tures used [10–12]. subjects: St Mercurius (1st zone)
 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110 93

Some complementary research data from the follow- Ultraviolet fluorescence photographs were taken
ing paintings (in that they contributed positive sup- using a SINAR 4× 5’’ camera and an illuminating
port to the general conclusions) have been added: system of 2 ×4 ultra-violet lamps, on a KODAK
The Lion of Judah (2nd zone) E-100S, 100 ASA daylight colour reversible film. In
St Theodore of Tyre (1st zone) front of the camera lens a KODAK 2E WRATTEN
Christ enthroned (1st zone) gelatine filter was placed.
At this stage, it would be useful to identify certain Colour measurements were taken on homoge-
fundamental particularities of the traditional Byzan- neous spot areas of 4 mm in diameter, under diffuse
tine technique of painting, in order to render com- reflectance and d/8° geometry, using a MINOLTA
prehensible the research methodology, as well as the CM-2022 portable spectrophotometer supplied with
expected results stemming from each application. a memory card for data storage, a CM A32 zero
In Byzantine wall paintings, colour layers are calibration box and a CM-1 colour data software.
superimposed one above the other. Colour measurement representation was realised on
– Proplasmos, the name given to the ground a) reflectance percentage (R%) spectra, and b) a*,
colour layer, is considered to be the basis for all the b*, CIELAB (1976) colour diagrams. Colour slides
discernible overlaying gradations. were digitised using a SCANVIEW Scanmate 5000
– Outlines (grapsimata) and highlights scanner, supplied with the 3.33 colour quartet ac-
( fotismata) are normally painted over the ground quisition and processing software.
layer (proplasmos).
– Drawing and sense of depth are rendered 3.2.2. Sampling analysis
through shadings (outlines and tinnings), which are Special attention was paid so that the sampling
mixtures of proplasmos and a black pigment (car- was representative of the work under consideration
bon black). and to avoid further irreversible damage. Once the
– Highlights are rendered when a white pigment samples were mounted in polyesteric transparent
(lime white) is mixed with another pigment, itself resin, they were ground and polished, a procedure
the chief constituent component of the ground performed by a STRUERS PLANOPOL-V machine,
colour layer (proplasmos), though added in a re- in such a way that their cross-sections provided all
duced proportion. The number of highlights usually related information about the existing stratigraphy.
varies. Cross-sections were observed under a LEICA DM
RXP research polarising microscope equipped with
a quartz halogen and an ultra-violet excitation light
3.2. Methodologies source, as well as an automatic photography device.
Raman spectra were recorded on the micro-sam-
3.2.1. Non-destructive analysis ples’ cross-section, using a RENISHAW system
Colour slides were taken in the visible area of the 1000 Raman spectrophotometer, comprising an
spectrum using a SINAR 4×5’’ camera with two OLYMPUS BH-2 imaging microscope. The spec-
electronic flashtubes 5 500 K on a KODAK EPR trophotometer was equipped with a grating
64 ASA daylight, code no. 6117, colour reversible monochromator and a charge-coupled device (CCD)
film. Moreover, for the requirements of the subse- Peltier-cooled detector (576× 384 pixels). The inci-
quent digitising procedure, the paintings were pho- dent laser excitation was provided by an air-cooled
tographed together with reference tables comprising HeNe laser source, operating at 632.8 nm. The
square-shaped (2 cm× 2 cm) and uniformly painted power at the exit of a ×50 objective lens varied
colour patches. These tables were prepared in the from 1 to 3 mW, depending upon the stability of
laboratory with a wide range of pigments, fre- identified pigments. Spectra were recorded with a
quently used in Byzantine iconography, and an egg resolution of 4 cm − 1, a collection time of 30 s and
yolk binder. Their colorimetric co-ordinates were an accumulation of ten scans. In order to avoid the
measured and well known. undesirable Rayleigh scattering, two notch-filters
Infrared reflectograms were taken using a HAMA- were used to block the laser line at 100 cm − 1. Pure
MATSU 2400-03D precision TV infrared camera silica served to calibrate the instrument.
equipped with a Vidicon infrared tube and a mFTIR spectra were obtained using a Biorad FTS-
NIKON F1,2/50 mm lens, in front of which an 45A FTIR spectrophotometer, connected to an
interference filter at 1 800 nm was placed. Image UMA 500 microscope and equipped with a mer-
capture and digitisation procedure was controlled cury-cadmium-telluride detector (MCT), cooled by
through special software. liquid nitrogen. In order to analyse the different
94 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110

components in every colour layer of the collected because of pentimenti in the final over-painting
samples, a small amount of each layer was removed (figures 1c, d and 2a, b).
from the sample using a micro-scalpel and then placed The detection of these pentimenti, in two cases
on the surface of a freshly prepared KBr pellet. For only, confirms and reveals the quality of absolute
each spectrum 250 consecutive scans were recorded certainty and virtuosity of design found in the work
with a resolution of 4 cm − 1 using a (× 15) objective of this great artist, who could produce richly drawn
lens. The sample collection area was adjusted with the compositions without the need of further correction.
upper aperture of the microscope. All spectra were The anatomical features of the figures are exposed
collected in the transmission mode and converted in detail, owing to the use of carbon black in the final
afterwards in absorbance spectra. As background, the outlines. This is a pigment that displays characteristi-
spectrum of the KBr pellet was used. cally high absorption in the infrared area of the
Elementary analysis was carried out through the spectrum. In the faces and hands, notwithstanding
use of scanning electron microscopy and electron that the rendering of final outlines follows a specific
probe micro analysis with energy dispersive analysis and systematic mode – with plain lines and a free
(ISI-SS40 SEM, accelerating voltage =20 kV. Sam- brushstroke movement – (according to the Macedo-
ples were mounted on aluminium stubs with carbon nian technique), an extraordinary variety of expres-
tape, and sputter coated with carbon for EDS analysis sion is achieved, thanks to Panselinos’ technique and
(EDAX PV9900 X-ray analyzer). Both methods can inspiration (figure 2 ). The rendering of colour-grada-
provide a qualitative and quantitative analysis of tion in the garments follows standard procedures;
inorganic substances. Plots describe precisely the however, the variety of positions, figure-movement
elementary distribution of the sample under and compositional patterns is impressive.
examination.
X-ray diffraction analysis was performed through 3.3.1.2. Pigment identification. Owing to the char-
the use of a GANTOLFI chamber with a Cu tube. acteristic absorption of certain pigments in the in-
frared area of the spectrum and thanks to the simple
3.3. Experimental results and discussion way in which the mixtures have been used, in many
cases pigment identification is discernible. Indeed, in
The experimental procedure, involving a combina- some cases, gradations (i.e. shadings and highlights)
tion of non-destructive and sampling techniques, has, stand out with considerable clarity. This identification
according to the methodology outlined above, yielded was made possible through the recognition of some
the following observations, which are presented be- grey levels, corresponding to the characteristic ab-
low. sorption of the detected pigments, in the infrared
reflectograms (table I).
The following are two characteristic examples.
3.3.1. Infrared reflectography In the detail of the reproduction of The Entrance
of the Mother of God (figure 3a, b):
3.3.1.1. Initial drawing – changes. Final drawing. – in the inner vestment, the low absorption of the
A number of initial drawing incisions in the garments, ground tone and tinning layers indicates the use of
as required by the fresco technique, became apparent. caput mortuum; owing to the addition of carbon
They are very limited and do not entirely cover the black, the outlines appear intensively dark;
folds of the clothes, while they are completely absent – high absorption in the gradations of grey high-
in the faces. The freedom of the final brush-strokes lights indicates the use of carbon black, while the
is nevertheless remarkable, because they do not fol- difference in tones is proportionate to the wt % of
low the discernible initial incisions (figure 1a). lime white;
It seems plausible to assume the existence of an – high absorption in the highlights of the red folds
original brush-stroke drawing, otherwise there would suggests the use of minium.
be considerable difficulty in creating art so complex In the detail of the reproduction of St Artemius
without some kind of initial drawing guide. It is, (figure 3c, d):
however, impossible to detect one by infrared reflec- The gradations of the red garment, visible in the
tography, either because of the low absorption of the corresponding reflectogram, are highly representa-
pigments used or as a result of the coincidence of the tive. Differences in the grey levels, as noted in the
final outlines with the initial drawing. infrared reflectogram, were confirmed by optical
Nevertheless, there are very few cases where the microscopy and mRaman spectroscopy (figure 4 ) and
original drawing is revealed and can be traced, corresponded to the use of:
 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110 95

– red ochre in the ground tone; 3.3.1.3. Current state of preservation. Pigment
– cinnabar in the first light; alteration: the alteration of minium was detected in
– minium in the second light (in areas where a number of places (figure 3c, d). In wall paintings,
minium was chemically altered, it revealed a near- minium alteration is caused by the effects of light
black hue); and occurs gradually with the passage of time. At
– carbon black in the outlines; first, it appears to be grey, but in its final stage of
– limonite over the sleeve indicated by the very alteration it turns a chocolate brown [14]. The
low absorption. chemical alteration consists of the transformation of

Figure 1. a) IR reflectogram. Detail of the reproduction The Entrance of the Mother of God. b) Photography in the visible
area of the same detail. c) IR reflectogram. Detail of the reproduction The Resurrection. d) Photography in the visible area
of the same detail.
96 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110

Figure 2. a) IR reflectogram. Detail of the reproduction Christ enthroned. b) Photography in the visible area of the same
detail. c) IR reflectogram. Detail of the reproduction The Resurrection. d) Photography in the visible area of the same
detail. e) IR reflectogram. Detail of the reproduction St Artemius. f) Photography in the visible area of the same detail.

(Pb3O4) to lead dioxide (PbO2). In the Protaton’s hair, in the representation of the Lion of Judah
frescoes, the first stage of alteration can be ob- (figure 5a), reveals damage in that area. Later at-
served. tempts to restore the paint, details of which are
Past attempts at restoration: several instances of discernible in the visible image (figure 5b), have
intervention during later restoration have been been made.
detected. This has been confirmed by related observations
The absence of paint at the left side of Christ’s on the microstratigraphic cross-sections through op-
 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110 97

tical microscopy. A sample taken from an area, – In certain scenes of the first zone, multiple
where the original painting is preserved (figure 5c), spots were revealed in UV fluorescence. A thorough
showed an irregular thickness and an expanded in situ observation identified them as wax spots
diffusion of the pigments into the underlying lime ejected at the rotation of the chandelier, a ritual
layer. However, where later intervention occurred performed during certain services (figures 6c and
(figure 5d), colour layers were more even in their 7a).
thickness and they did not exhibit pigment – There were indications of later surface inter-
diffusion. ventions, which consisted of consolidation work.
Owing to the relative degree of absorption, there During these interventions over time, different or-
is considerable difference in the painting of St ganic materials were used, each one bearing differ-
Artemius’ right eye as compared with that of the left ent fluorescent hue. In contrast with the rest of the
(figure 2e, f). This could be noted in the infrared scenes (figure 7c), an extremely intense blue–green
reflectogram, which also provided proof for later fluorescence could be observed especially in the
intervention. scenes of the first zone (figures 6c and 7a).

3.3.2. Ultra-violet fluorescence photography 3.3.2.2. Pigment identification. Unlike naked-eye

visibility, when the painted images were examined
3.3.2.1. Delimitation and the study of surface after UV-fluorescence photography, it became possi-
damage and later interventions. Through ultra-violet ble to identify the following pigments, all of which
fluorescence photography, detectable surface dam- exhibited an individual fluorescence of high inten-
age received greater clarity. The damage was com- sity. In fact, this characteristic fluorescence, in the
prised of the following. visible area of the spectrum, was observed when the
– Superficial salt deposits caused by the flow of a reference tables of pigments used in Byzantine
water/cement mixture suspended during the con- iconography were excited by a UV light source.
struction of a cement support in the upper part of Precisely, as far as the pigments identified in the
the fourth zone as part of the restoration of the Protaton frescoes are concerned, the characteristic
church’s ceiling (figure 6a – c). fluorescent hue was:
– Extended flakings and detachments of the – minium: orange fluorescence
colour layers caused by blockage and the conse- – azurite: deep blue fluorescence
quent obstruction to the evaporation of natural – limonite: yellow–brown fluorescence
humidity. This was the result of the addition of a – warm ochre: light brown–orange fluorescence
cement coat in the lower part of the inner walls of – red ochre: deep red fluorescence
the first zone (figures 6c and 7a). The following is a typical example of pigment
identification.
Table I. Correspondence of infrared reflectogram grey lev- Detail of the representation of The Entrance of
els with the characteristic absorption of the detected pig- the Mother of God (figure 7c, d):
ments. – minium: identified in the highlights of Mary’s
robe; the vestment and high priestly mitre of the
Pigment Grey level in the
reflectogram
prophet Zachariah; St Anne’s robe; the fourth vir-
gin’s garments; the flames of the virgins’ lamps; the
Carbon black black virgins’ shoes;
(very high absorption) – azurite: identified in the highlights of Mary’s
Azurite dark grey peplum; the vestment of St Anne; the blue vestment
(high absorption) of the prophet Zachariah; and in some of the other
Minium altered in composition dark grey virgins’ garments;
(PbO2) (high absorption) – limonite: identified in the haloes; the first vir-
Minium(Pb3O4) grey gin’s cloth; the head-covering of the third virgin
(medium absorption) from the left; and in the scarf of the virgin behind St
Green earth grey
(medium absorption)
Joacheim;
Red ochre, caput mortuum, light grey – warm ochre: identified in the cloth of the virgin
limonite (low absorption) at the edge.
Cinnabar, calcium white white These indications have been confirmed by optical
(minimum absorption) microscopy and mRaman spectroscopic analysis
(figure 4 ).
98 Sister Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110

Figure 3. a) IR reflectogram. Detail of the reproduction The Entrance of the Mother of God. b) Photography in the visible
area of the same detail. c) IR reflectogram. Detail of the reproduction St Artemius. d) Photography in the visible area of
the same detail.

3.3.3. Colour measurements In figure 8b, c spectra of the reflectance per cent
In order to study the current state of colour (R%) versus wavelength l, which refer to the sur-
preservation in Panselinos’ paintings in relation to face colour layers of the selected measurement posi-
their original hue, saturation and lightness levels, a tions, are indicated. From these spectra, hue,
cleaning test was performed in eight places of the saturation and lightness levels were calculated and
representation of The Entrance of the Mother of represented in the L*a*b* CIE 1976 colour system.
God (in limited areas of 5 × 5 cm, where, through Table II comprises the numerical results of the
difference values in hue (DH*ab), saturation (DC*ab)
optical observation, the painting was judged to be
and lightness (DL*) in the L*a*b* CIE 1976 both
homogeneous). The areas cleaned are shown in before and after the cleaning tests. From this table it
figure 8a. In these areas the cleaning was carried out is obvious that the only important and significant
in two stages. The first stage consisted of a simple difference is limited to the DL* value, while the
mechanical dry cleaning (rubbing with cotton wool), DH*ab and DC*ab difference values remain low and are
while in the second distilled water was also used. virtually negligible. This means that, with respect to
Colour measurements were taken before and after their hue and saturation, the colours have remained
each stage of the cleaning. practically unchanged, given the fact that layers of
 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110 99

whatever kind of impurity, when deposited on the representation of colour stimuli, but, given the fact
frescoes’ surface, are colourless. They have, there- that the one is a mathematical conversion of the
fore, deprived the wall paintings only of their origi- other, they lead to the same conclusions.
nal luminosity (brightness). It is quite possible that The measurements carried out on 12 scenes (table
this is essentially a layer of smoke and perhaps some III) are represented fully in the a*b* diagram of
dust (the latter to a much lesser degree). L*a*b* CIE 1976 colour system, where the posi-
This conclusion has lent verity to the systematic tions of the pure colours R G B Y M and C are also
measuring of colour in a large number of the Prota- exhibited, in relation to the white that the iconogra-
ton thematic scene representations, thus permitting pher used and was therefore measured in the Prota-
the determination of the colour palette, at least ton frescoes (figure 9a). The measurement of this
where hue and saturation are concerned. white, the colour of the underlying lime layer, was
For the present study, two systems of measure- taken from the white ground layer (proplasmos) of
ment and colour representation were used: L*a*b* St Mercurius’ shield (measurement no. 6 in the
CIE 1976 and x y Y CIE 1931. In substance, the representation of St Mercurius). Its co-ordinates are
difference between the two systems lies in their x = 0.3466, y= 0.3614 and = 39.59 in xyY (CIE

Figure 4. a) Raman spectra of the pigments detected in Panselinos’ frescoes: azurite, green earth, limonite, warm ochre. b)
Raman spectra of the pigments detected in Panselinos’ frescoes: red ochre, cinnabar, minium (and minium alterated). c)
Raman spectra of the pigments detected in Panselinos’ frescoes: carbon black, calcium carbonate, caput mortuum.
100 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110

Figure 4 Continued.

1931) and a*= 1.13, b* =13.89, L* = 69.17 in The complementarity of colour combinations, sys-
L*a*b* CIE 1976 colour systems, for the standard tematically observed in Panselinos’ painting, has
illuminant D65. also been studied and represented in a x, y chro-
By observing all of the colour measurements, we maticity diagram. Some characteristic examples are:
can appreciate that Panselinos’ palette covers the – the complementarity of the yellow–orange
entire gamut of colour hues. However, colour mea- haloes and the blue background;
surements were small in number in the yellow– – the complementarity of yellow–orange and the
green and green–cyan areas. Green, grey – green and blue tones and of green and purple tones of gar-
olive hues, chiefly observed in the Baptism and some ments (figure 10a).
other scenes, were placed on the diagram a* b*, in Finally, a grouping of several colour combinations
the vicinity of the white – grey hues of low saturation presenting approximately the same hue – such as
levels. purple, blue, yellow, yellow–brown, orange, yel-
A general observation that may be made on Pro- low–olive, green and red garments and flesh tones
taton wall paintings is that the total range of colours was created, and a record was made of the spectrum
used by Panselinos are of low saturation levels. of reflectance per cent (R%) versus wavelength l, in
Chroma C* values are located far from the the visible area (400–700 nm).
boundary curve corresponding to the maximum A comparative study of these spectra yielded se-
colour density for the lightness value L*=40, which cure evidence that the iconographer has used the
is proximate to the average lightness value of the same pigments or the same mixture of pigments in
total of measurements (L*$40 9 5) (figure 9b). each case.
S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110 101

Figure 4 Continued.
102 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110

The form of these spectra remains unchanged Purple tones: caput mortuum + calcium white +/
every time; the observed fluctuations in the Y axis carbon black.
are certainly due to the varying dilution of these Green tones: green earth+ calcium white+ / car-
pigments with either white or black pigments (for bon black.
example figure 10b). Yellow tones: limonite+ calcium white.
From the same spectra the simplicity of the colour Red tones: red ochre+ carbon black. The combi-
mixtures used can be detected. One pigment is usu- nation of red pigments for rendering the resulting
ally mixed with white (lime white), a feature that colour gradations (highlights–shadings) is character-
was cross-checked by determining the composition istic and concrete (table IV).
of the colour layers through analytical techniques. Grey–blue tones (background+ garments):
The systematic and representative collection of white + black+ azurite (characteristic maximum at
reflectance spectra (R% versus wavelength) in a : 480–500 nm) only where azurite remained
present.
large number of colour tones of different colour
As far as complex mixtures are concerned, the
hues, which describe chromatically the adjacent 12
shape of the spectrum curve is subject to fluctua-
scenes under examination, led to the following con- tions according to the maximum values, which cor-
clusions with respect to their qualitative chemical respond to the dominant colour hue.
composition. These conclusions were verified Olive, olive–yellow, brown–olive, brown–yel-
through mRaman spectroscopy pigment low, brown tones mixture: green earth + limonite+
identification. red ochre + white +black.

Figure 5. a) IR reflectogram. Detail of the reproduction The Lion of Judah. b) Photography in the visible area of the same
detail. c) Cross-section of a sample taken from the left side of Christ’s hair (original painting). d) Cross-section of a sample
taken from the right side of Christ’s hair (later intervention).
 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110 103

Figure 6. a) UV-fluorescence photography. Detail of the reproduction The Resurrection. b) Photography in the visible area
of the same detail. c) UV-fluorescence photography. Detail of the reproduction St Theodore of Tyre. d) Photography in the
visible area of the same detail.
104 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110

Figure 7. a) UV-fluorescence photography. Detail of the reproduction St Mercurius. b) Photography in the visible area of
the same detail. c) UV-fluorescence photography. Detail of the reproduction The Entrance of the Mother of God.
d) Photography in the visible area of the same detail.
 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110 105

Figure 8. a) Detail of the reproduction The Entrance of the Mother of God. The eight square-shaped areas indicate
locations where a cleaning test was performed. b) Reflectance per cent R% versus wavelength l diagrams before and after
the first and second stage of the cleaning test. c) Reflectance per cent R% versus wavelength l diagrams before and after
the first and second stage of the cleaning test.

Table II. Numerical results of the difference values (absolute mean values and maximum values) in hue (DH*
ab), saturation
(DC*ab) and lightness (DL*) in the L*a*b* CIE 1976, both before and after the experimental cleaning tests (superficial
rubbing with cotton wool). First stage: dry cleaning, second stage: with distilled water.

Representation of Dab DC*

ab DL* DECMC
The Entrance of
the Mother of God

After the first stage 0.4 (max DH*ab =0.7) 0.8 (max DC*ab = 3.2) 4.5 (max DL*=5.9) 4.5 (max DCMC = 5.9)
After the second stage 1.0 (max DH*ab =1.1) 2.8 (max DC*ab = 10.5) 10.4 (max DL*= 13.1) 10.7 (max DCMC = 14.4)

– Olive tones: higher content of green earth. technique of applying the colour layers was used: a
– Olive–yellow yellow – brown tones: higher con- fresco and a secco.
tent of limonite. The following factors provide proof for the use of
– Brown tones: green earth + limonite + red a fresco technique: a) the high content of chaff in
ochre. the lime wash; b) the morphology of the initial
Ground tones of faces and hands: green earth+ colour layers and their intense diffusion in the
limonite. preparation layer; c) the presence of horizontal join-
Flesh tones: limonite+red ochre + white. ings between the succeeding zones (in vertical
positioning).
3.3.4. Optical microscopy Their shape identifies a descending work progres-
sion from ceiling to floor. The independent parts of
3.3.4.1. Panselinos’ painting technique. An obser- every horizontal zone describing adjacent thematic
vation through the optical microscope of samples’ ensembles do not include vertical joinings or lime
stratigraphy led to the conclusion that a mixed wash coatings between neighbouring scenes, as wit-
106 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110

Figure 8 Continued.
 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110 107

nessed by the fact that the paint work is completed

while the lime wash is still wet.
The wide use of lime as white pigment over the
entire painted area, the obvious, firm resistance of
the colour layers, as well as the visual impression of
the painting (the observed luminosity, the volume of
highlights and the semi-transparency of the colour
layers), support the aforementioned observations.
The wall’s preparation layer (lime mortar), made
with lime wash, consists of two main ingredients,
calcium carbonate CaCO3 (lime) and chaff. This
was verified through X-ray diffraction and optical
microscopy. Silicium, in the form of quartz and
albite (a silica compound, NaAlSi3O8), low in con-
tent, was detected as a secondary component. The
high content of lime wash in hydrophile chaff re-
duced the progress of the standard carbonisation
procedure, thus allowing the application of a fresco
technique and assuring the total coherence and co-
hesion of the paintings’ structure. This explains the
high resistance and stability of Panselinos’ frescoes.
Time limitations, owing to the quick drying of the
surface of the wet lime wash, clearly did not allow
the artist to complete the painting using a fresco
technique. In the overlying colour layers, the pres-
ence of calcium carbonate grains provides evidence
that pigments are mixed with lime, the binding
material that also serves as a white pigment.
However, in the stratigraphy of the red garments
(highlights with cinnabar and minium) no mixing
with calcium carbonate was observed figure 11a.
This fact, together with the characteristic ultra-violet
fluorescence emitted by the red colour layers (high-
lights), when observed through the optical micro-

Table III. Colour measurements carried out on 12 scenes.

Subject Number of
measurements

Resurrection 33
St Artemius 26 Figure 9. a) a*b* diagram (L*a*b* CIE 1976) of the total
Baptism 28 of colour measurements. b) a*b* diagram (L*a*b* CIE
Christ enthroned 16 1976) of the total of colour measurements. On the same
St. John Damaskinos 19 diagram the boundary curve corresponding to the maxi-
The Entrance of the Mother of 37 mum colour density (chroma Cab * ) for lightness value
God L*= 37.8 is plotted.
Dormition 34
St Luke 36
St Mercurius 28 scope, under an ultra-violet light, gave proof once
Virgin 14 more of the presence of an organic vehicle. That this
St Theodore the Recruit 14 organic component is egg-based was verified by
St Theodore of Tyre 10 FTIR spectroscopic analysis (figure 11b). The distin-
guished peaks at 2 992 and 2 855 cm − 1, which are
108 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110

Figure 11. a) Cross-section of a characteristic highlight

sample taken from the folds of the red virgin’s garment in
the reproduction The Entrance of the Mother of God.
b) FTIR spectrum of the binder taken on the second red
layer. The red pigment, which is also present in the same
layer (cinnabar), does not generate any characteristic
peaks within this area.

attributed to methyl groups, as well as the intensive

Figure 10. a) x, y chromaticity diagram CIE 1931 of peak at 1 660 cm − 1 – characteristic of amide car-
complemental colour combinations. b) Reflectance per bonyl groups (CONH) – followed by a shoulder
cent R% versus wavelength l of resembling colour hues at 1 735 cm − 1 – characteristic of ester groups –
spectra coming from several purple garments.
yielded evidence of the presence of an egg-based
binder. The small shift of this shoulder towards
lower wavenumber values can be due to the natural
Table IV. Combination of red pigments for rendering the ageing of the egg yolk [15]. The presence of an
resulting colour gradations (highlights–shadings). egg-based binder in the red highlights was also
verified through gas chromatography (GC-MS) anal-
Nature of pigment Rendered colour gradation ysis of the same sample (Colombini, unpublished
data).
Red ochre ground tone (proplasmos)
Ced ochre+carbon black outlines
Cinnabar 1st light 3.3.4.2. Panselinos palette. According to all ex-
Minium 2nd light (highlight) perimentation results derived from the spectroscopic
methods used, such as mRaman (figure 4 ), mFTIR,
 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110 109

and elementary microanalysis by scanning electron – Thanks to the a fresco technique and the nature
microscopy, Panselinos’ colour palette consists of the of the pigments used, the Protaton frescoes, generally
following ten traditionally used natural pigments: speaking are in a good condition.
– calcium carbonate or lime white (CaCO3) – Subsequent fillings in the paintings have been
– carbon black (C) detected.
– limonite (Fe2O3·H2O) – Localised surface damage was due mainly to
– red ochre (Fe2O3), warm ochre (Fe2O3·H2O) external phenomena, such as: a) deposits of salts; b)
and caput mortuum (Fe2O3) extended flaking and detachments of colour layers,
– green earth (celadonite) in the lower part of the inner walls of the first zone;
– azurite (2CuCO3·Cu(OH)2) c) local chemical transformation of the minium pig-
– cinnabar (HgS) and minium (Pb3O4), are used ment; d) multiple wax spots in several scenes of the
in exceptional cases (red highlights) first zone caused by wax ejection during the rotation
of the chandelier; e) later interventions of surface
consolidation, using organic materials especially in
4. Conclusions the representations of the first zone.
These results and conclusions, arising from the
The collective results of experimentation on the application of diagnostic procedures to the Protaton
Protaton frescoes have yielded the following frescos, are offered as a contribution towards the
conclusions. accurate restoration of their colours and as a lesson
Knowledge of Panselinos’ painting technique con- for the prevention of future detrimental activity.
stitutes a valuable lesson in Byzantine iconography
for contemporary icon painters.
Its main characteristics are the following. References
– The application of a mixed technique: initial
colour layers (ground tones – proplasmos) are di- [1] Achimastou-Potamianou M., Helliniki Techni: Byzan-
rectly applied onto a wet limewash (a fresco) while, tines Toichografies, Ekdotiki Athinon, 1994.
in the superimposed colour layers, lime is used as a [2] Chatzifoti I.M., Makedoniki Scholi, I scholi tou
binding agent. In the case of red highlights, an egg Panselinou (1290 – 1320), Ethniko Idruma Neotitos, 1995.
vehicle substitutes the lime binder. [3] Dionysiou Ek Fourna Ton Agraphon, Ermineia tis
– Thanks to his virtuosity of design, Panselinos Zografikis Technis, A Papadopoulou-Kerameos, Petrou-
combines simplicity of drawing with an enormous polis, 1909.
diversity of dynamic expression. Changes in the [4] Winfield D.C., Middle and Later Byzantine Wall
initial drawing are extremely rare. Paintings Methods, Dumbarton Oaks Papers, 1968.
– Although Panselinos’ palette includes a wide [5] Chryssoulakis Y., Chassery J.M., The Application of
variety of colour combinations, these combinations Physicochemical Methods of Analysis and Image Process-
ing Techniques to the Painted Works of Art, Athens,
are simple and repetitive. The multiple colour combi- 1989.
nations frequently preserve a complementarity in
[6] Kosolapov A.I., Infrared Reflectography of Paintings,
pairs, whereby total chromatic harmony is achieved. Theoritical and Experimental Research, ICOM Commit-
Low saturation levels are the result of the important tee for Conservation, 6th Triennial Meeting, Ottawa,
dilution of pigments, which, in the case of simple 1981.
mixtures, produces a transparency in the colour [7] Van Ansperen De Boer, Reflectography of paintings
layers. using a IR Vidicon television system, Stud. Conserv. 14
– A standard technique is used for the structure (1969) 96 – 118.
of the colour layers (simple stratigraphy, simple [8] Delaney J.K., Metzger C., Wamsley E., Fletcher C.,
mixtures) using an extremely limited number of Examination of the visibility of underdrawing lines as a
pigments. The pigments used are: lime white, carbon function of wavelength, 10th Triennial meeting, Washing-
ton DC, USA, 1993, pp. 15 – 19.
black, azurite, green earth, caput mortuum, yellow,
[9] Wamsley E., Metzger C., Delaney J.K., Fletcher C.,
warm and red ochres, cinnabar and minium. Improved visualization of underdrawing with solid state
The following remarks can be made concerning detectors operating in the infrared, Stud. Conserv. 39
the current state of Panselinos’ Protaton frescoes. (1994) 217 – 231.
– The surface layer of impurities, being achro- [10] Billmeyer F.W., Saltzman M., Principles of Colour
matic, could be removed by mechanical cleaning, Technology, 2nd ed., Wiley, Chichester, 1981.
thereby allowing exposure of the colours’ initial [11] Trouve A., La mésure de la couleur, Principes, Tech-
luminosity. nique et Produits du marché, AFNOR-CETIM, 1991.
110 S. Daniilia et al. / J. Cult. Heritage 1 (2000) 91 – 110

[12] Wyszecki G., Stiles W.S., Colour Science: Concepts [14] Gettens R.G., Stout G.L., Painting Materials. A
and Methods, Quantitative Data and Formulae, 2nd ed., Short Encyclopaedia, Dover, New York, 1966.
Wiley, New York, 1982.
[13] Elgina-Ter Haar G., On the use of the electron [15] Meilunas R.J., Bentsen J.G., Steinberg A., Analysis of
microprobe in analysis of cross-sections of paint samples, aged paint binders by FTIR spectroscopy, Stud. Conserv.
Stud. Conserv. 16 (1971) 41–45. 35 (1990) 33.