12 International Congress on the Deterioration and Conservation of ...

<strong>12</strong> th <strong>International</strong> <strong>Congress</strong> on the Deterioration and
Conservation of Stone
Tuesday 23 October 20<strong>12</strong>
Oral Presentations—Testing and (Re-)Evaluation;
Documentation
Session VI: 2:00 – 4:00

THE USE OF A WIRELESS SENSOR NETWORK FOR HIGH-RESOLUTION
ENVIRONMENTAL MONITORING OF STONE MONUMENTS IN CONTEXT
WITH INVESTIGATION OF SALT WEATHERING – EXEMPLIFIED FOR
ROCK-CUT MONUMENTS IN PETRA / JORDAN
Kurt Heinrichs 1 , Rafig Azzam 1 and Markus Krüger 2
1 Department of Engineering Geology and Hydrogeology, RWTH Aachen University
2 TTI GmbH – TGU Smartmote, Stuttgart
Abstract
Salt weathering is known as a major cause of damage on stone monuments
worldwide. Despite many years of intensive research, processes of salt weathering are
still rather poorly understood. The overall aim of the research project ‘petraSalt’ is to
improve knowledge of salt weathering on stone monuments. The project addresses salt
weathering processes on stone monuments under real-time / real-scale conditions. It
aims at reliable information on characteristic interrelations between stone properties,
environmental influences (climate, salt load), salt weathering processes and the
development of weathering damage. In order to ensure findings of high transferability,
the rock-cut monuments of Petra / Jordan were selected for studies, since stone type and
spectra of monument exposure regimes, environmental influences, salt loading and
weathering damage are representative for a multitude of stone monuments worldwide.
Achievement of the project´s aims requires temporal and spatial high-resolution
monitoring of environmental conditions affecting the monuments and acting as driving
forces for salt weathering processes. An autonomously operating wireless sensor
network (WSN) is applied as an innovative technology that can make this contribution.
Methodological approach - with focus on the WSN-system - and results of the
‘petraSalt’ research project are presented.
Keywords: stone monuments, salt weathering, weathering damage, wireless sensor
network, environmental monitoring, impedance spectroscopy
1. Introduction
Stone monuments represent an important part of world heritage. All stone
monuments are affected by weathering. Findings obtained from in-situ investigation,
laboratory analyses and weathering simulation during the last decades have revealed salt
weathering as a major cause of damage. Despite many years of intensive research,
processes and mechanisms of salt weathering still cannot be explained satisfactorily.
This is due to the heterogeneity of the systems “stone”, “salt” and “environmental
influences” and the complexity of their dynamic interaction. From experts´ point of
view, better understanding of salt weathering deserves further comprehensive in-situ
investigation jointly addressing active salt weathering processes and controlling factors.
The ‘petraSalt’ research project – funded by DFG / Deutsche Forschungsgemeinschaft
(German Research Foundation) – takes this approach. The project addresses salt
weathering processes on stone monuments under real-time / real-scale conditions.

In order to ensure high transferability of methodological approach and findings, the
rock-cut monuments of the ancient city of Petra in Jordan were selected for studies,
since stone type (sandstone) and spectra of monument exposure regimes, environmental
influences, salt loading and weathering damage are representative for a multitude of
stone monuments worldwide. The Nabataean city of Petra with its more than 800
monuments carved from bedrocks about 2,000 years ago (tombs, sanctuaries, places of
worship) represents outstanding world cultural heritage (Figs. 1 and 2). In 1985
UNESCO inscribed Petra on the list of World Heritage. In 2007 Petra was elected to
represent one of the “New Seven Wonders of the World”.
Figure 1. Monastery (No. 462) Figure 2. Corinthian Tomb (No. 766)
The project aims at weathering models that reliably depict characteristic
interdependencies between stone properties, monument exposure regimes,
environmental influences, salt loading and salt weathering damage. These models are
expected to allow reliable rating and interpretation of aggressiveness and damage
potential of the salt weathering regimes considering their variability under range of
lithology, monument exposure, environmental conditions and time (Fig. 3).
Monument
exposure
scenario,
lithology /
stone
properties
Climate
Salt load
Time and space
Salt weathering
Stone deterioration
Weathering progression
Figure 3. Dynamic weathering situation to be faced
Weathering
damage

In addition to precise assessment of characteristic monument exposure regimes,
weathering damage and salt load regimes, achievement of the project´s aims requires
temporal and spatial high-resolution monitoring of environmental conditions affecting
the monuments and acting on the salt load as driving forces for salt weathering
processes, considering diurnal, seasonal and spatial variation. An autonomously
operating wireless sensor network (WSN) is applied as technology that can make this
contribution. This innovative technology increasingly finds application in geosciences.
In the framework of the ‘petraSalt’ research project it has entered the field of systematic
salt weathering research.
2. Working program
The studies focus on the two main sandstone types which show considerable
differences in their petrographic properties and in their weathering behavior as known
from previous own research work in Petra (Heinrichs 2005, 2008):
• multicolored, fine-grained sandstone, Umm Ishrin Sandstone Formation – middle
part / Cambrian,
• whitish, medium-grained sandstone, Disi Sandstone Formation / Ordovician.
Representative monuments were selected for studies, considering different
architecture, monument exposure regimes and extent of weathering damage. The
working program comprises five main activities:
• assessment of characteristic monument exposure scenarios (location, dimension and
architectural composition / geometry of the monuments; orientation of stone
surfaces, rain impact, water run-off situation, conditions of insolation and stone
surface heating / cooling, etc.),
• assessment of weathering damage (type, extent, spatial distribution and progression
of weathering damage),
• continuous high-resolution monitoring of micro-environmental conditions affecting
the monuments (temperature, humidity, insolation, rain, wind),
• assessment of characteristic salt load regimes (type, concentration and spatial
distribution of salts / salt mixtures; salt weathering profiles),
• integral evaluation.
These activities are explained in the following chapters with focus on the use of the
wireless sensor network system for environmental monitoring.
3. Assessment of monument exposure regimes
As basic contribution to the assessment of characteristic monument exposure
regimes, 3D terrestrial laser scanning (TLS) was made. In a first step, laser scanning
was performed from longer distances with a resolution in the range between 2 and 6 cm,
aiming at overview 3D sceneries of the monuments and their surroundings. For detailed
investigation of the monuments, close-range scanning was carried out with a resolution
in the range between 6 and 15 mm (Figs. 4 and 5). Evaluation of the laser scanning data
by means of 3D models has provided precise information on dimension and stone
surface exposition of the monuments under investigation (Heinrichs and Nguyen 2011).
In addition to thorough visual inspection of the monuments, supplementary
measurements and available information on the Nabataean architecture in Petra, the 3D
models of the monuments served as basis for the assessment of the monuments´ original
architectural composition including ground section (Figs. 6 and 7).

Figure 4. Bab el-Siq Triclinium (No. 34),
Ordovician sandstone
Figure 5. Bab el-Siq Triclinium, 3D model
(front view) derived from laser scanning
This step of evaluation is required for precise quantification of weathering damage
on the monuments, especially loss of stone material. Furthermore, the 3D models of the
monuments were used for the calculation of rock mass that was removed for the creation
of the monument façades. Considering the huge mass of rock material removed,
unweathered condition of the monuments´ rocks can be postulated for the initial phase
of exposure. Thus, salt weathering processes can be clearly attributed to a period of
approximately 2000 years. A next step of investigation has addressed the rainfall and
water run-off situation at the monuments, considering rain as an important source of salt
loading. Stone surface exposed to rain was identified. Statistical information on
direction and inclination of rainfall was gathered allowing a differentiated calculation of
rain water input at the monuments in dependence on orientation of their stone surfaces
as well as the assessment of spatial variation of stone surfaces exposed to rain in the
course of weathering progression. Furthermore, those parts of the monuments affected
by water run-off during or after rain were mapped, considering short and long,
respectively light and heavy rainfall (Fig. 8).
Figure 6. Bab el-Siq Triclinium, monument
plan derived from reconstruction of the
original architecture by use of the 3D model of
the monument, approximately 700 m 3 of rock
removed for creation of the monument
Figure 7. Bab el-Siq Triclinium, ground
section, 1: chamber with three graves, 2:
triclinium chamber with two graves (a), 3:
chamber, 4: chamber with four graves

Figure 8. Bab el-Siq Triclinium, 3D image with documentation of water run-off, dark
gray areas with hatching: water run-off after light / short rain, light gray areas with
hatching: additional areas by water run-off after heavy / long rain
Measurements by means of infrared-thermography are performed for the
assessment of the heating and cooling behaviour of the monument façades as further
important contribution to identification of characteristic monument exposure scenarios.
The measurements provide detailed information on parameters such as minimum,
maximum, and average stone surface temperatures, insolation periods, variation of stone
surface temperature, heating and cooling rates etc., considering diurnal and seasonal
variation (Figs. 9 and 10).
Stone surface temperature (ºC)
50
45
40
35
30
25
20
15
10
5
0
03:30:00 AM
Bab el-Siq Triclinium (No. 34), 7 July 2011
04:30:00 AM
05:30:00 AM
06:30:00 AM
07:30:00 AM
08:30:00 AM
09:30:00 AM
surface heating
(~ <strong>12</strong> hours)
Figure 9. Bab el-Siq Triclinium, stone surface
temperatures in high summer (example),
façade exposed to NW
10:30:00 AM
11:30:00 AM
<strong>12</strong>:30:00 PM
insolation (~ 6 hours)
Standard time (local time / summer time - 1 hour )
Maximum stone surface temperature Minimum stone surface temperature
Average stone surface temperature
01:30:00 PM
02:30:00 PM
03:30:00 PM
04:30:00 PM
surface
cooling
(~ <strong>12</strong>
hours)
05:30:00 PM
06:30:00 PM
Figure 10. Bab el-Siq Triclinium, average
diurnal variation of stone surface temperatures
(vertical profile) in high summer (example)

4. Assessment of weathering damage
Weathering damage on the rock-cut monuments is assessed by mapping of
weathering forms (apparent weathering phenomena at cm to m-scale) according to type
and intensity. A classification scheme of weathering forms has been developed as basis
for detailed mapping, specially tailored to use for the Petra monuments. It is
accompanied by representative photo-documentation. Evaluation comprises:
• maps of weathering forms (separately for loss of stone material, deposits, current
detachment of stone material, cracks),
• quantitative evaluation of all weathering forms and their combinations,
• identification of characteristic spectra and sequences of weathering forms
(development of weathering damage, weathering progression) in consideration of
the sandstone types and monument exposure regimes under investigation,
• rating of weathering damage by use of a correlation scheme “weathering forms –
damage categories” (maps and quantitative evaluation of damage categories,
calculation of damage index).
5. Environmental monitoring by use of a wireless sensor network (WSN)
An autonomously operating wireless sensor network system (WSN) has been
developed in the framework of the ‘petraSalt’ research project for the temporal and
spatial high-resolution environmental monitoring of the monuments selected for detailed
investigation. Considerable advantages of this innovative technology compared to
conventional measuring techniques are miniaturisation of sensors (high quality, low cost
sensors), possibility of complex sensor setups, high-resolution monitoring, high
durability of the system, little maintenance efforts, and remote access to system and data
by the end-user. The WSN system is composed of several independent wireless nodes
(motes), linked to each other by a short range radio communication link (2.4 GHz),
hence building a wireless sensor network (Fig. 11).
Drill hole
INSIDE STONE STONE SURFACE
Sensor mote
↓
data collection from
all sensors
Cylinder with
6 sets of sensors in
various depths
MEASURING UNIT
Set of sensors
Figure 11. Configuration of the wireless sensor network, measuring unit with sets of sensors and
sensor mote for data collection

Regarding measurements inside the stone, a cylinder was developed that is
equipped with six sets of sensors, arranged in separate chambers in various depths (1, 3,
6, 9, 14, 19 cm). Each set includes sensors for the measurement of temperature,
humidity and electrical impedance. The cylinder is inserted in drill holes (depth: 20 cm,
diameter: 3.7 cm, dry drilling). A special spreading technique with closed cell foam
made from silicone ensures proper sealing of the six measuring chambers against each
other as well as tight and dense connection between cylinder and stone substrate.
Reversibility of this technique will ensure that the measuring units can be easily
removed at the end of the project without harming the stone substrate.
Exterior system components of the measuring units – densely connected with the
cylinder inside the drill hole – comprise sensor mote with rain-proof enclosure, long-life
batteries, processor board, radio module and sensors for the measurement of air
temperature, air humidity, stone surface temperature and light (insolation).
In addition, several measuring units have been equipped with wind and rain sensors.
In this way, a measuring unit can be equipped with 24 sensors at maximum. The sensor
motes act as data collector and are programmed to collect data from all sensors in
periodic intervals of 15 to 30 minutes over a period of at least one year.
Additional elements of the wireless sensor network system are solar powered
gateways, which relay the measurement data to a long-distance network (GSM) for
remote access, and a database to save data for later retrieval and optional postprocessing.
The database is a MySQL database located in Germany. The possibility of
remote access to the data allows their continuous evaluation (Fig. <strong>12</strong>).
Sensor
motes
Gateways in Petra
remote
access
per
internet
Webserver, database (Germany)
Figure <strong>12</strong>. Configuration of the wireless sensor network, data transfer from the sensor motes via
gateways per internet to remote webserver / database
All the hardware is optimized to work under harsh environmental conditions.
Different kinds of sensors could be attached to a wireless mote simultaneously that is
various MEMS (Microelectromechanical systems) sensors with digital output, e.g. for
the acquisition of temperature, humidity, solar radiation etc.
data
commands

Additionally analog sensors like electric impedance sensors are connectable by
using especially developed electric circuits for the signal conditioning. This modular
concept allows for customization and optimization for specific monitoring objectives as
they are apparent in Petra. While sufficient techniques and sensors are available for the
measurement of temperature and relative humidity of the air, the determination of
moisture content of natural stone is of high complexity. All porous materials could
contain various amounts of moisture that depends on the specific material, especially its
chemical composition and porosity, and on the surrounding environment and exposition.
The hygroscopic moisture content of a material is driven by absorption and desorption
functions of both relative humidity and temperature and represents the amount of water
absorbed from the air at a standing environmental level. A lot of attempts have been
made to monitor the equilibrium moisture content by just measuring the relative
humidity in boreholes. Although such measurements can give some indications on
material moisture and moisture change over time, the determination of absolute moisture
by this method could be source of severe error, because the equilibrium moisture content
strongly depends on hygroscopic substances inside the material, to which many chloride
and hydroxide salts belong.
The determination of moisture content becomes even more complex if the
overhygroscopic area is concerned. In excess of the hygroscopic moisture, additionally
free (or capillary) moisture is filling the pores of the material. The free moisture content
could vary significantly between different stone varieties because of their different
permeability and porosity and therefore moisture carrying capacities as well as different
salt content. This is why in addition to the measurement of the equilibrium moisture
content impedance measurements will be conducted simultaneously. Due to the given
restrictions of the measurement hardware it was decided to use a two-point electrode
setup, although it is known that electrode effects (e.g. double layer) influence the
measurement results (Barsoukov 2005). The electrodes used are made from conductive
silicone and are pressed onto the stone surface inside the small boreholes. In general the
measured impedance is dependent on the amount of water and the salt concentration
within the water as well as temperature. The parameters influence the real and imaginary
part of the impedance in a different way. Therefore, only measuring both parts in
combination with the equilibrium moisture content and temperature allows conclusions
about the moisture and salt concentration in the stone. However, the impedance value is
also frequency dependent. Thus, the impedance sensor board is capable of assessing the
impedance over a frequency range of 10 Hz to 100 kHz. Impedances starting from a few
Ohms up to approximately 15 MΩ can be measured. For Petra the system used for the
impedance measurements was set to measure logarithmic full-frequency sweeps from 10
Hz to 100 kHz with 10 points per order of magnitude.
6. Assessment of salt load regimes
The drill cores obtained from installation of the wireless sensor network system are
taken for segment-wise ionic analysis of soluble salts (0 - 0,5 cm, 0,5 – 1,5 cm, 1,5 – 2,5
cm,...18,5 – 19,5 cm). This ensures detailed information on the spatial distribution of the
salt / salt mixtures (depth profiles). The ECOS program (ECOS - ‘Environmental
Control of Salt Damage)’will be used as an expert thermodynamic model for the
prediction of the crystallization behaviour of the salts / salt mixtures (salt phases,
amount of salt, salt volume) under changing climate conditions (Price 2000).

Application will be made via RUNSALT (user interface to the ECOS-program)
(Bionda 2002-2005). In addition, efflorescences on the stone surface of the monuments
under investigation, detritus (loose material) from the base of the monuments and water
(precipitation, water run-off) are planned to be included in geochemical analysis in
consideration of composition, concentration and origin of salts.
7. Integral evaluation
All results obtained from measurements with the wireless sensor network,
supplementary in-situ investigation and laboratory analysis will get combined for overall
comprehensive information on salt weathering and its effects on the Petra rock-cut
monuments. The integral evaluation will comprise:
• statistically reliable information on correlations between monument architecture,
monument exposure scenarios, climatic regimes (stone surface, stone interior,
diurnal and seasonal variation) and effects of salt weathering (state / development of
weathering damage),
• characterization and quantification of salt loading (type and extent of salt
contamination, spatial distribution of salts, main zones of salt accumulation,
individual salts / salt mixtures, degree of pore space filling etc.) in consideration of
the monument exposure regimes, salt uptake mechanisms and states of weathering /
weathering progression,
• detailed numerical analysis of potential salt dissolution / salt crystallization cycles
according to frequency and depth, based on joint evaluation of salt loading and
microclimate acting on it,
• comparison of salt loading (depth profiles) and potential salt dissolution /
crystallization cycles (frequency, depth) in consideration of seasonal variation
• correlation of salt regimes and weathering phenomena in the course of weathering
progression,
• comparative characterization of the two sandstone varieties under investigation with
respect to their weathering behavior / susceptibility to salt weathering in
consideration of the differences in their petrographic properties,
• conclusive rating and interpretation of aggressiveness and damage potential of salt
regimes in consideration of stone type, monument exposure characteristics,
environmental influences, salt loading and salt weathering processes.
8. Conclusions
The overall aim of the ‘petraSalt’ research project is to improve knowledge of salt
weathering on stone monuments. For reasons of representativeness the rock-cut
monuments of the ancient city of Petra in Jordan were selected for the investigation. The
methodological approach of the research project combines the application of a wireless
sensor network system (WSN) as an innovative technology for high-resolution temporal
and spatial monument environmental monitoring with systematic, well-directed
laboratory analysis and supplementary in-situ investigation of the selected Petra
monuments.
The wireless sensor network will provide an extraordinary information output for
the analysis of interrelations between exterior and stone interior environmental
conditions at the stone monuments considering diurnal, seasonal and depth-dependent
variation.

Moreover, it will provide the basis for detailed analysis of depth-dependent salt
crystallization-dissolution processes in dependence upon monument exposure
characteristics, salt load, environmental influences, state of weathering and stone
properties. In this way the application of the wireless sensor network will make an
important contribution to rating of aggressiveness and damage potential of salt regimes
in consideration of their variability.
9. References
Barsoukov, V.S. and Macdonald, J.R. 2005. Impedance spectroscopy. Theory,
experiment, and applications, 2nd edn.: John Wiley & Sons.
Bionda, D. 2002-2005. RUNSALT computer program. http://science.sdf-eu.org/runsalt/.
Heinrichs, K. 2005. ‘Diagnose der Verwitterungsschäden an den Felsmonumenten der
antiken Stadt Petra / Jordanien’. Ph.D. dissertation – Geological Institute, RWTH
Aachen University.
Heinrichs, K. 2008. ‘Diagnosis of weathering damage on rock-cut monuments in Petra /
Jordan’. Environmental Geology, Special Issue "Monument future: climate change,
air pollution, decay and conservation - The Wolf-Dieter-Grimm Volume" 56 (3): 643-
675.
Heinrichs, K. and Nguyen, H.T. 2011. ‘3D terrestrial laser scanning of rock-cut
monuments in Petra / Jordan’. Mitteilungen zur Ingenieurgeologie und
Hydrogeologie, Lehrstuhl für Ingenieurgeologie und Hydrogeologie, RWTH Aachen,
104: 27-37.
Price, C. 2000. An expert chemical model for determining the environmental conditions
needed to prevent salt damage in porous materials. Protection and Conservation of
the European Cultural Heritage, Research report No 11, Archetype Publications Ltd,
London (UK).
10. Acknowledgements
The authors give their thanks to DFG – Deutsche Forschungsgemeinschaft (German
Research Foundation) for project funding. Furthermore, the authors would like to
express their gratitude to the representatives of the Department of Antiquities of Jordan
(DOA) and the Petra Development & Tourism Region Authority / Petra Archaeological
Park & Cultural Heritage Office (PDTRA / PAP) for cooperation, discussion, advice and
all support of the research project in Petra. Further thanks go to SENSIRION who
supported with their sensors the development of the wireless sensor network.

DAMAGE ASSESSMENT OF FERRUGINOUS SANDSTONE
BY X-RAY TOMOGRAPHY-
THE “VIRGIN TOWER” OF ZICHEM (BELGIUM)
Hilde De Clercq 1 , Roald Hayen 1 , Veerle Cnudde 2 , Matthieu Boone 2 , Michiel Dusar 3
1 Royal Institute for Cultural Heritage, Brussels, Belgium
2 University of Ghent, Centre for X-ray tomography, Ghent, Belgium
3 Geological Survey of Belgium, Brussels, Belgium
Abstract
The ferruginous sandstone of the gothic “Virgin Tower” is suffering from a specific
biological deterioration process triggered by perforating activities of mason bees. The
damage due to these perforations causes extensive loss of material, so that a durable
conservation of such degraded stone blocs becomes questionable.
In order to evaluate the conservation possibilities of stone blocs damaged by
perforating mason bees, an investigation of the internal structure by means of X-ray
tomography was carried out. This investigation revealed that the cumulative effect of the
digging work by multiple generations of mason bees may result in networks of
perforations. Bioturbated sandstones were found to be most suitable for attack by mason
bees because of morphological and geometrical compatibility between the original
layered burrowings by marine organisms and those by the mason bees. As a conclusion
the conservation is not recommended of sandstone blocs for which the load bearing
capacity is endangered by the branched and layered perforations.
Keywords: Diestian ferruginous sandstone, mason bee, perforation, X-ray tomography
1. Introduction
Hageland is an area to the north-east of Brussels in Belgium, characterised by a
particular landscape of rolling hills in steeply-sloped longitudinal rows. They are
underlain by marine, medium-coarse, poorly sorted glauconitic sands of the Diest
formation, of Miocene age and protected from erosion by alteration of part of the
glauconite and limonitic cementation in the oxidation zone following the hill
morphology (Gullentops 1996, Bos 1990 and Regionaal Landschap Noord-Hageland
2007). The lithification thus is a late diagenetic process related to pedogenesis, probably
still ongoing but slowly under the present climate. The resulting ‘Diestian’ ferruginous
sandstone is of heterogeneous composition and variable quality. Quartz dominates apart
from around 30% of glauconite. Clays are present as films around the sand grains and
some infillings, preventing effective goethite cementation. The total porosity attains
25% while the density 2050 kg/m³. The average compressive strength of quality stones
is <strong>12</strong> N/mm² with a sound velocity of 2420 m/s. However, alteration processes occurring
between or even within apparently similar stones may lead to lowering of the quality to
8 N/mm² and a sound velocity of 1370 m/s (Van Campenhout 2009).
The rusty brown Diestian ferruginous sandstone is a local building material, but
very typical for the Hageland and linking its architectural heritage to the landscape
(Gulinck 1949, Doperé 2003 and Dusar 2009) (Figure 1). Diestian sandstones were most

enowned for constructing gothic buildings, although quarrying for vernacular use and
for restoration works continued till the early 20th century. Upon extracting the still soft
quarry-stone material, more attention was paid to their format and regularity than to the
ultimate durability of the material. As a consequence, exposure to natural weathering
processes often causes important damage. Although it is generally assumed that reserves
must be vast, today no potential quarry locations with high quality Diestian sandstone
are known. Hence, restoration of several iconic monuments built with this ferruginous
sandstone for which the degradation pattern implies replacement is a difficult topic.
The Virgin Tower (Dutch “Maagdentoren”) is a dungeon erected in 1387 in the
alluvial plain of the river Demer at the outskirts of the former city of Zichem. The round
tower with a height of 26 m and wall thickness of 4 m at ground level was built in
ferruginous sandstone apparently coming from a single source. Renovation works did
not affect the dressed sandstone masonry which till recently has nearly kept its original
dimension. Although considered as heritage of exceptional value for the Flemish region
(Doperé 2002 and Doperé 2003), the tower suffered from neglect and partially collapsed
in 2006 (Fig. 2). Following urgent consolidation works (Breda 2008), a conservation
and restoration strategy had to be determined (Vanderauwera 2008). The ferruginous
sandstone masonry of the Virgin Tower suffers from severe deterioration phenomena for
which adequate restoration methods have to be established. Generally, the ferruginous
sandstone is characterised by the presence of a superficial black crust serving as natural
protection of the stone. This limonite rich skin tends to fall off, leaving behind a weak
surface to natural exposure (ICOMOS ISCS 2008).
Figure 1. Representation of ferruginous sandstone of the formation of Diest as building stone
in Belgian monuments, concentrated in the Hageland, a landscape corresponding to the geological
outcrop of the Diest sand and characterised by hills underlain by Diestian sandstone.

Figure 2. Virgin Tower at Zichem, an abandoned medieval four stage dungeon erected in the
14th century, after its partial collapse in 2006 due to loading by roof breakdown, important soil
deposit and long-lasting water infiltration. The structure has since been consolidated in its present
form and protected against further infiltration. The present study deals with the standing walls not
affected by the collapse.
Moreover, both ferruginous sandstone and mortar are suffering from a specific
biologic deterioration process caused by the burrowing of mason bees identified as
Osmia cornuta (Grootaert s.d.) (Figure 3). Since a few decennia perforations by mason
bees have become pervasive and are probably linked to the changing environment since
the surrounding area gained the status of protected natural landscape. Female mason
bees, active in early spring, use especially the weaker parts of the ferruginous sandstone
and the mortar to lay their eggs. Every year, a new nest is made by these solitary bees.
Although perforations are observed all around the tower, especially the south-east side
is heavily affected. Locally damage by multiple perforations results in crumbling and
erosion of the stony material for which a durable conservation becomes uncertain.
Figure 3. Detail of ferruginous sandstone masonry severely damaged by the perforations of
mason bees.

Proper conservation strategies cannot be defined from a visual inspection of the
degree of damage caused by the mason bees. Stone blocs presenting few perforations at
the surface may be completely crumbled inside due to a branched structure of the
perforations. In order to evaluate the conservation/restoration possibilities of stones
damaged by mason bees and to interpret the load bearing capacity, knowledge of the
internal structure is necessary. Important questions to be answered are the depth range of
visible perforations and their three dimensional structure inside the stone.
A visualisation of the internal structure can be obtained by means of X-ray
tomography (CT-scan). X-ray tomography is a “non-destructive” three dimensional
imaging technique (Cnudde 2006 and Masschaele 2007). A well-known application is
the medical computer tomography (CT) for visualisation of the internal structure of the
body. Nowadays, there is a growing interest for X-ray tomography in scientific and
industrial fields. The strong technical evolution of X-ray sources and detectors enables a
resolution of 1 µm, depending on the size of the object (approximately 1:1000th of the
maximum diameter of the object). An object of 10 mm diameter can hence be scanned
with a resolution of 10 µm.
A radiographic image of an object is a « shadow image », based on the intensity of
absorption of X-rays radiated through the material. Through rotation of the object,
radiographic images at different orientations are taken. These images are combined into
a 3D volumetric reconstruction of the sample, which can then be divided into 2D grey
scale slices. The grey scale is a measure of the absorbed X-rays in the voxel (volumetric
pixel element) which depends as such on the density of the material. The output is a
virtual 3D image of the scanned object obtained after manipulation of the scans. The
principle of micro-CT is presented in Figure 4.
Figure 4. Principle of micro CT.
2. Experimental investigation
The tested samples are:
• Drilled cores of ferruginous sandstone (6 cm diameter);
• Two subsamples with a diameter of 4 mm taking out of the drilled cores. One
sample was taken from the inside of the core ("bulk" sample), while the other
sample perpendicular to the surface of the stone which contains the surface
crust ("crust" sample);
• Two ferruginous sandstone blocks, “small” (<strong>12</strong> x 8 x 6 cm) and ”large” (14
x 10 x 10 cm) lifted from the masonry; and,

• One mortar sample (15 x 8 x 5 cm).
The experimental conditions of the X-ray tomography are:
• X-ray source: Feinfocus FXE-160.51, directional head, 150 kV, focal spot: 10
µm
• Hardware filter: 1.05 mm Cu and 3.00 mm Al
• X-ray detector: Varian 2520V PaxScan a-Si flat panel, CsI, 1880x1496 pixels,
<strong>12</strong>7 µm pixel pitch
• Distance source-detector: 830 mm
• Distance source-object: 248.5 mm
• Enlargement : 3.34x resulting in a voxel pitch of 38 µm (drilled cores) and
1.15x resulting in a voxel pitch of 109 µm (stone and mortar blocs)
The scans of the subsamples were analysed by means of Morpho+, the 3D analysis
software package developed at the Centre for X-ray tomography of the Ghent University
(Vlassenbroeck 2007).
3. Results and discussion
3.1 Drillcores
Figure 5 illustrates the 2D-scan through a drillcore lifted from a moderately
damaged stone block. The burrow and nest of the mason bee can be clearly recognised
down to a depth of 5 cm below the wall surface. For the yearly digging of a new nest,
pathways created by old burrows are preferentially used. Existing burrows inside the
stony material may get filled with loosened ferruginous sandstone fragments. Figure 6
shows such an old burrow filled with stone fragments and then covered with a layer of
fresh clay, probably picked up by the mason bee from the alluvial plain nearby. Figure 7
illustrates the backfilling of unused parts of older burrows. Within the same sample,
bioturbations, created by burrowing organisms and giving rise to a textural feature of
many ferruginous sandstones inherited from times of sedimentation, can be clearly
visualised in the CT-scan. These bioturbations also appear darker coloured in the CTscan
indicative for their lower density and hence mechanical properties, but less so than
the refilled perforations produced recently by the mason bees. It is remarkable that the
original burrows of the marine organisms and the ones dug by the mason bees have
quite comparable dimensions.
3.2 Subsamples
The CT-scans of the two subsamples with a diameter of 4 mm were analysed by
means of Morpho+ software to evaluate their internal porosity changes. In this study a
significant different porosity was detected between the "crust" sample and the "bulk"
sample. When analysing the partial porosity according to the Z-axis (going from the
surface into the interior of the stone) a lower value was obtained at the surface compared
to the interior part, due to the formation of a dense surface crust. The results are
discussed in detail elsewhere (Cnudde 2011).

Figure 5. Photograph of a drillcore surface showing a recent mason bee burrow covered by a
fleece (left) and corresponding CT-scan (right) (core diameter 6 cm).
Figure 6. Visualisation of a mason bees’ burrow filling: ferruginous sandstone fragments
loosened during digging of a younger burrow and covered with a protective layer of clay (dark
grey) (length of the image: approximately 6 mm).

Figure 7. Reconstruction of CT-scans of a the outer part of a ferruginous sandstone drillcore.
The black arrow marks a bioturbation dug by burrowing animals during the sedimentation
process of the ferruginous sandstone. These bioturbations are characterised by an almost
uncemented sand filling, hence of lower density and darker coloured, surrounded by a more
strongly cemented rim, hence denser and light coloured. The red arrow marks a perforation filled
with ferruginous sandstone fragments produced during digging of a burrow by a new generation
of mason bees. The burrow filled with sandstone fragments is darker coloured indicative for its
lower density (size of the image: 4 cm).
3.3 Ferruginous sandstone block samples
Perforations are preferentially dug in the weaker parts of the ferruginous sandstone.
Ferruginous sandstones may present a succession of weakly and more strongly
cemented zones following concentrations of goethite cement in mostly flattened nodular
forms (corresponding to Liesegang rings) (Gullentops 1996 and Bos 1990). Bioturbated
sandstones are built up of weakly cemented burrows surrounded by a more strongly
cemented rim. These burrows provide easy access to mason bees. Figure 8 illustrates
how such a bioturbated sandstone has become moderately damaged by the perforations
of mason bees. The corresponding CT-scan illustrates the correspondence between the
original bioturbations and the digging by the mason bees. As the bioturbations have
layered sequences, which is the case for the stone block presented in figure 8, or
represent entire sandstone blocs, its load bearing capacity is in danger. Therefore, the
conservation of such a stone, although visually moderately damaged, cannot be
recommended.
The CT-scan of the small tested stone block, presented in figure 9 (left), illustrates
some holes dug by mason bees. Through image treatment of the scans of successive 2Dslices
the structure of the dug holes present in the stone block could be reconstructed, as
shown in figure 9 (right). This reconstruction shows the branched structure of the dug
holes which is as such lowering the mechanical properties of the stone. Also in this case,
replacement is recommended.
Similar results were obtained for the mortar sample presented in figure 3 for which
conservation is neither recommended.

Figure 8. Bioturbated ferruginous sandstone (length: 14 cm ; width: 10 cm) heavily attacked
by mason bees (recent perforations marked by light grey clay lining), extracted from the
Maagdentoren (left). The corresponding CT-scan (right) shows the original burrows by sea
organisms (worm-like faint grey bioturbations), most clearly visible in the upper part of the scan,
and similar zones which attracted the mason bees whose perforations are visible in the middle and
the lower part of the scan.
Figure 9. Ferruginous sandstone block (size: <strong>12</strong>x6x8 cm), CT-scan of the block (left) and
negative-density 3D-reconstruction showing the internal distribution and connectivity of the
burrows dug by mason bees (right).
4. Conclusions
The ferruginous sandstone of the Maagdentoren in Zichem is prone to a specific
biological deterioration process triggered by mason bees of the species Osmia cornuta.
Since a few decennia, the activity of mason bees has become worrisome and is probably
linked to environmental protection measures for the surrounding landscape. These

solitary bees preferentially attack the weaker parts of the sandstone and the mortar to lay
their eggs, most so at the south-east side of the tower. Locally the damage has caused
crumbling of the stone materials, so that a durable conservation of such biologically
damaged stones becomes questionable.
In order to evaluate the conservation possibilities of stone and mortar perforated by
mason bees, an investigation of the internal structure of damaged building materials was
carried out by means of X-ray tomography (CT-scan).
From the obtained scans, it could be concluded that the burrowing depth by each
mason bee is approximately 4 to 5 cm. Preferentially, for the digging of a new nest,
existing old burrows are followed which get either prolongated or backfilled with
loosened ferruginous fragments.
Original burrowings by marine organisms have led to bioturbated sandstones with
heterogeneous texture and durability. Weakly cemented zones occurring throughout
these sandstone blocks are preferentially followed by the mason bee burrows. Insofar
the bioturbations affected by the mason bees have a layered structure, the load bearing
capacity of the stone is in danger. The conservation of such a stone, even when visually
only moderately damaged, is not recommended.
Moreover, through image processing of the CT data the structure of the perforations
could be reconstructed revealing a branched nature, due to progressive attack by
generations of mason bees, apparently returning to their site of birth. When these
branched perforations progress inside the stone material, its mechanical properties are
endangered and hence also in this case, replacement is recommended.
From this investigation it could be concluded that this very fast biological
destruction process of the 14th century tower needs urgently to be stopped. In situ
destruction of the mason bees, which are irreplaceable pollinators, by means of spraying
is from an ecological point of view not acceptable. Rather it is recommended to dissuade
the mason bees from returning each spring to the stone structure by providing alternative
housing in the near neighbourhood, as is already commercially organised in some fruit
producing regions. This way, a symbiosis could be realised between natural landscape
protection and heritage conservation. If such an approach would prove inadequate, one
should opt for allowing the former agricultural activities in the near neighbourhood of
the tower, or let the tower gradually decay over time as nature takes control.
5. Acknowledgements.
We thank AROHM for their financial support and interest. The Fund for Scientific
Research-Flanders (FWO) is acknowledged for the post-doc grant to V. Cnudde.
6. References
Bos, K., Gullentops, F. 1990. 'IJzerzandsteen als bouwsteen in en rond het
Hageland', in Bulletin de la Société belge de Géologie, 99: 131-151.
Breda, K., 2008. 'De Maagdentoren. Instandhoudingswerken afgerond' , in
ICOMOS contact, 21(1): 17-18.
Cnudde, V., Dewanckele, J., Boone, M., De Kock, T., Dusar, M., De Ceukelaire, M.,
De Clercq, H. and Hayen, R. 2011. 'High-resolution X-ray CT for 3D petrography of
Belgian ironsandstone with crust formation', in Microscopy Research and Technique,
74(11): 1006-1017.

Cnudde, V., Masschaele, B., Dierick, M., Vlassenbroeck, J., Van Hoorebeke, L. and
Jacobs, P. 2006. 'Recent progress in X-ray CT as a Geosciences Tool', in Applied
Geochemistry, 21(5): 826-832.
Doperé, F. 2003. 'Het gebruik van kalkzandsteen en ijzerzandsteen als technische
basis voor het ontstaan en de ontwikkeling van de gotische architectuur in het
hertogdom Brabant.' Bijdragen tot de geschiedenis, Antwerpen 86: 347-371.
Doperé, F., Klinckaert, J., Minnen, B. and Van der Eyken, M. 2003. Bouwen met
ijzerzandsteen in de Demerstreek (Brabantse Bouwmeesters), Provincie Vlaams-Brabant,
Leuven.
Doperé, F. and Corens, K. 2002. 'Huizen in torens. De Zichemse Maagdentoren en
andere donjons'. Openbaar Kunstbezit in Vlaanderen.
Dusar, M., Dreesen, R. and De Naeyer A. 2009. 'Natuursteen in Vlaanderen,
versteend verleden'. Kluwer Renovatie & Restauratie.
Grootaert, P., Personal comm., Dept. Entomology, Royal Belgian Institute of
Natural Sciences.
Gullentops, F. 1996. 'IJzerzandsteen', in Delfstoffen in Vlaanderen. Ministerie van
de Vlaamse Gemeenschap, Dept. EWBL, Admin. Natuurlijke Rijkdommen en Energie :
88-89.
Gulinck, M. 1949. 'Oude natuurlijke bouwmaterialen in Laag- en Midden-België'.
Technisch-Wetenschappelijk Tijdschrift, 18(2): 25-32.
ICOMOS-ISCS (<strong>International</strong> Scientific Committee for Stone) (Vergès-Belmin, V.
ed.) 2008. 'Illustrated glossary on stone deterioration patterns'. Monuments and Sites
XV, ICOMOS, Paris.
Masschaele, B., Vlassenbroeck, J., Dierick, M., Cnudde, V., Van Hoorebeke, L. and
Jacobs, P. 2007. 'UGCT: New X-ray Radiography and Tomography Facility', in Nuclear
Instruments and Methods in Physics Research A., 580(1): 266-269.
Regionaal Landschap Noord-Hageland 2007. Groeves met een ijzer-sterk verhaal.
De ijzerzandsteengroeves van het Hageland, Regionaal Landschap Noord-Hageland,
Aarschot.
Van Campenhout, D. 2009. Verleden en toekomst van de exploitatie van
ijzerzandsteen uit het Diestiaan als bouwsteen, een geologische studie. M.Sc. thesis,
KU Leuven, Dept. Earth & Environmental Sciences.
Vanderauwera, M., 2008. 'Verslag van de uitstap naar Zichem', in ICOMOS contact
21(1): 15-16.
Vlassenbroeck, J., Dierick, M., Masschaele, B., Cnudde, V., Van Hoorebeke, L. and
Jacobs, P. 2007. 'Software tools for quantification of X-ray microtomography at the
UGCT', in Nuclear Instruments and Methods in Physics Research Section A:
Accelerators, Spectrometers, Detectors and Associated Equipment, 580(1): 442-445.

LEARNING FROM THE PAST – A REVIEW OF TREATMENTS CARRIED OUT TO CLUNCH IN 1985
AND FURTHER FIELD AND LABORATORY INVESTIGATIONS INTO THE DECAY MECHANISMS
David Odgers 1 , Trevor Proudfoot 2 ,
1 Odgers Conservation Consultants, Somerset, UK
2 Cliveden Conservation, Berkshire, UK
Abstract
Clunch is a sedimentary limestone (chalk) from the Lower Cretaceous period. When first quarried it is greenishgrey
in colour (due to the presence of glauconite) and contains an abnormal amount of small fragments of the fossil
bivalve ‘Inoceramus’. It has a significant porosity (more than 30%) and most of these pores are very fine as they
consist of the gaps between very small grains (about 0.01mm diameter). The stone, which was widely used in the
south and east of England since the 14 th century, can be subject to significant and severe deterioration principally
through a characteristic lamination of the surface; the decay mechanism has never been fully explained.
In Chapter 11 (Case Studies) of the first edition of Practical Building Conservation: Vol 1 Stone Masonry
(1988), John Ashurst described a series of comparative treatments that were undertaken on the clunch of the west
elevation of the south stable block at Woburn Abbey in August 1985. One-half of this elevation had been dressed
back to behind the decay zone and the other had not. The same treatments were applied to both types of surface; they
included an alkoxy silane consolidant (Brethane), hydrophobic treatment (microcrystalline wax, boiled linseed oil)
and lime treatment (lime poultice, limewater and lime-casein shelter coat).
It is rare that the opportunity arises to carry out a complete assessment of such a variety of treatments but in
2010, a thorough investigation of the treated elevation was carried out and this paper will describe the results and put
them into context with past, current and possible future treatments of the stone. The results were also used to support
further investigations into the decay mechanisms; these included simple field techniques such as permeability
measurements and decay mapping as well as laboratory analysis of core samples with EDX.
Keywords: Clunch, decay, treatment trials, permeability, options
1. Introduction
Chalk from the Upper Cretaceous beds has played a significant part in the historic architecture of Southern
England. Most chalk is soft and weathers easily but is still found in some building interiors. The harder beds within
the chalk are generally referred to as ‘clunch’. The scarcity of good building stone in the area meant that this was
extensively used for external walling as well as more decorative elements of carving and vaults. The decay of clunch
is generally more pronounced than other limestones with significant loss of surface and it has always been assumed
that this close-grained texture and fine pore structure of the stone has been a significant contributor to its decay.
There have been many interventions carried out over the years to mitigate the causes and deal with the effects
of decay. These have included the use of limewash (often as part of a maintenance programme), oils and waxes as
surface treatments. Historically repairs often involved the tooling back of the affected surface or the replacement of
the stone (sometime using a different stone) and more recently repairs using lime mortars, grouts and shelter coats
have been widely used.
In order to test these various approaches in the field, the Research and Technical Advisory Group of the
HBMCE (now English Heritage) undertook some trials at Woburn Abbey in 1985. Soon after their completion,
conclusions were drawn as to which had been the most successful. Although there have been some informal
inspections in the intervening years, there has been no detailed review of the trials. As part of a general review of the
treatment of stone at Woburn Abbey, an opportunity arose in 2010 to assess the trials carried out 25 years previously
and, at the same time, to look in more detail at the nature and mechanism of the decay of clunch.
The assessment was mostly site based and was set up to include detailed visual recording, simple techniques for
objective measurement (such as permeability tests and USB microscope) and backed up by some laboratory based
tests.
2. Geology and properties of the stone
Clunch exists as a well-marked bed within the Lower Chalk; it is a thin bed in Berkshire, continues through
Oxfordshire and Buckinghamshire and reaches between 15 – 25 feet (4.6 – 7.6 m) thickness through Bedfordshire,
Hertfordshire and Cambridge (Dimes 1990). Clunch was quarried in two distinct areas – Tottenhoe in Bedfordshire
and near Cambridge. When first quarried it is greenish-grey in colour (due to the presence of glauconite) and
contains an abnormal amount of small fragments of the fossil bivalve ‘Inoceramus’. This results in it having a gritty
feel which has led to the stone being incorrectly described as ‘sandy’. Its texture is compact and fine-grained and it is
usually worked as a freestone (i.e. the bedding plane is not obvious so the stone can be worked in any direction). In
this paper we are principally concerned with stone from the Tottenhoe quarry.

Examination of the Tottenhoe stone shows that much of it is generally of poor quality. It has up to 8% clay
content and a microporous structure. Results from the BRE classification show that it has a porosity of 31.4% and
water absorption of 14%. There is however a bed in the lower 4.7m succession at Tottenhoe (which has a quarry face
of 46m). This seam is 0.7m in height and is both coarser and appears to have less micro-porosity as well as being
slightly greyer in colour; it can be identified by the inclusion of phosphatic nodules (Sanderson 2010).
Photomicrographs of the two different types of Tottenhoe clunch are shown below (Figs 1 - 2). The images
represent 1.22mm wide areas of the stones. The clearer and more colourless parts are crystalline pieces of shell and
the darker indistinct material is lime mud. The Tottenhoe Grey (more durable) stone has a much smaller proportion
of mud. A difference in particle size of the larger fractions, between the two types is clear to see. The grey variety
is coarser but apparently the matrix is more compact and opaque, while the white shows a greater degree of bright
speckling due to micrite particles which seem to be less closely packed than those of the grey. Although at the
present magnification pores are not discernable (they are considerably smaller than the thickness of the section, and
thus obscured by overlapping grains), it would seem that the softness of the white variety could be attributed to less
contact adhesion between grains and consequent greater permeability.
Fig 1: Photomicrograph of Tottenhoe (grey) stone Fig 2: Photomicrograph of Tottenhoe (white) stone
3. Historic use of clunch stone
Although chalk has been used since Roman times for construction including the traditional rammed chalk walls
found in the East Anglian region, Clunch appears to have been used since the Norman period (eleventh century). It
particularly found favour because it was easy to work and ideal for carved detail.
The lack of transport meant that small local quarries provided much of the clunch used. However larger quarries
such as Tottenhoe were used to supply stone to important buildings such as Windsor Castle and Westminster Abbey.
Clunch stone was extensively used in the construction of many churches in the medieval period up until the
dissolution of the monasteries (1535-40) when all monastic building stopped. The demand for stone declined until
in the middle of the eighteenth century, when the revival of classical architecture, enabled Tottenhoe and other
quarries to flourish. For nearly a century, the stone was much in demand as wealthy landowners built or remodeled
houses (such as at Audley End, Woburn, Ashdown, Southill and Ashridge). The material continued in localised use
for more modest projects (for example the Swan Hotel in Bedford) up to the last quarter of the nineteenth century, by
which time its limitations had become apparent and the transport network had developed so that other stones and
building materials became accessible.
When first quarried, the clunch is very soft and prone to damage during transport. For this reason, it was
customary to leave a block of clunch to season over a winter or two. During this time, the ‘quarry-sap’ moved to the
surface and the stone became harder. As it dried out, it also became much more permeable to moisture to the extent
that any mortar applied to the face would dry out so quickly that effective bedding of the stone was compromised. In
order to overcome this, there are references to coating internal faces of ashlar blocks with materials (such as linseed
oil and even bitumen) that would reduce the suction of the stone.
4. Mechanisms of decay – current understanding and recent investigations
It has been well recorded that the decay of clunch can occur quite quickly. Archive records at Woburn suggest
that within 40 years of construction, some of the clunch stonework was being ‘dragged and cleaned’. The BRE tests
on the building limestones of the British isles showed that Tottenhoe clunch had the lowest durability rating of any
limestone tested although it was also noted that some of the stone after many years weathering acquired remarkable
toughness and resistance to deterioration.
4.1 Decay characteristics

Typically, clunch has been observed to deteriorate in the following ways:
a) Multi-planar cracking due to loads acting on inherent fissures and faults in the stone
b) Despite its wide acceptance as freestone, incorrect bedding can reduce the load-bearing capacity of clunch
which has a low shear resistance. Exposure to the weather of any fine fissures between bedding planes may
lead to rapid deterioration.
c) Discolouration through various factors including soiling, sulphation and accumulation of biogrowths.
d) Lamination of the surface to a depth of 2 - 5mm (Fig 3). Some stones are susceptible to lamination but
others immediately adjacent (which are subject to the same environmental conditions) appear to be
completely sound. Once lamination starts, it will tend to develop over the whole stone.
e) Once the surface has been lost through lamination, the underlying stone that is revealed has powdered as a
continuing part of the same decay process; only in a few cases does the surface revealed beneath a
lamination then stabilise and not shown further decay.
Fig 3: typical lamination of surface of clunch
4.2 Investigation of decay mechanisms
The decay mechanism of clunch has never been fully identified and explained. It has a significant porosity
(more than 30%) and most of these pores are very fine as they consist of the gaps between very small grains (about
0.01mm diameter). It has been suggested that the fundamental causes of decay of clunch (and particularly the type of
Tottenhoe stone used extensively at Woburn) ‘relate to moisture rhythms and thermal cycling and the microporous
nature of the stone’ (Ashurst 1991). The blocking of the surface pores and the consequent reduction of the
permeability might occur due to movements of minerals to the front of the stone or by the rubbing of the stone face
during the construction process (and so clogging up the pores with fine dust known as slurry ); it may also be
attributed to ‘quarry-sap’. Any or all of these are thought to lead to the observed laminations.
Like all stones, the decay of clunch will also depend on factors such as design, orientation, exposure, use and
previous treatment or repair. During the investigations at Woburn Abbey, two parts of the elevations of the North
Court were studied. These were the north facing courtyard elevation (Fig 4: E03) and the west facing external
elevation (Fig 4: E04).
Fig 4: Plan of North Court of Woburn Abbey
4.3 Decay mapping
There are great benefits in overlaying notations of decay and surface changes on elevation drawings or
photographs; patterns of decay become discernible and help direct the investigation. Decay mapping was carried out
to indicate areas of lamination, powdering and to record previous repairs that had involved either tooling off the
surface (to a depth of about 8mm) or replacement of the stone. This mapping revealed that the elevations could be
divided into four zones (Figs 5 and 6):

Zone A – At low level and consists almost entirely of stone where the original surface has been lost and decay
continues to underlying stone through spalling and powdering.
Zone B – Up to lintel level of ground floor windows; this has most of the stone showing either current or
historic (often tooled off) lamination but with only individual stones exhibiting on-going spalling.
Zone C – Up to just above cill level of upper window; in this zone, the stone is mostly sound but with some
individual stones having a laminated surface and localised decay beneath.
Zone D – Up to cornice level; this area is almost all sound and the more protected stone has original surface
intact. This is also the area where surface coatings (limewash) are found.
Fig 5: Elevation E22 of North Court Fig 6: Elevation E04 (external) of north court
The difference between the two elevations is that in the more exposed elevation (E04), zones C and D arte much
narrower with Zone B being predominant. This reflects the greater exposure of this west facing elevation to rainfall
and more regular regimes of wetting/drying and heating/cooling.
4.4 Permeability tests
A number of stones were selected for permeability measurements using a Karsten tube – the locations are shown
in Fig 7. This covered all of the zones identified in the condition survey as well as different types of face (e.g.
weathered, laminated, re-dressed) of individual stones.
Fig 7: Location of porosity measurements in Bay E22
No 1: Decayed stone from Zone A
No 2: Sound surface (with brown biogrowth) in
Zone B
No 3: Tooled back surface in Zone B – adjacent to
No 2
No 4: Sound surface at top of Zone B
No 5: Sound surface in Zone C
No 7: Sound surface with coating in Zone D
No 8: Laminated section of stone in Zone C
Although conclusions are somewhat limited by the number of tests carried out, some useful interpretations can
be made:
- Most stones show a typical permeability curve with the absorption of water tailing off with time.
- No 1 (decayed stone with no intact surface) is the most permeable.
- Nos 2 and 3 show little difference suggesting that a tooled back surface eventually establishes a similar low
permeability as stones where the surface remains intact.
- The stones at higher level (for example No 7 with surface coating) have higher permeability than others
lower down; this suggests that the mechanism by which pore blocking takes place has not occurred to the
same extent – either due to the protected location of the stone or the presence of the surface coating.
4.5 Microbiology
Where the surface is intact, the stone generally has a brown and white patchy appearance with considerable
cultures of algae and lichen. The two elevations of the North courtyard were inspected by Professor Mark Seaward
and his comments can be summarised as follows:

- The extent and nature of colonisation has been dictated by the properties of the underlying stone; these vary
even on a stone-by-stone basis.
- Various treatments of the stone in the past (particularly with limewash) have influenced the colonisation.
- The extent of the grey lichen growth suggests that there has been a treatment over all the surfaces but it
manifests itself stronger on certain stones perhaps due to mineralogical differences.
- The brown surface (examined using a Raman spectroscope) showed that there were organic residues
present, but it seems certain that these were not metabolic products (i.e. derived from plant or fungal
growths) but related to organic treatments of the stonework.
- No calcium oxalate was detected, so this confirms there were no lichens (at least growing in these sample
spots in recent decades).
- Lichen biodeterioration was currently minimal or indeed insignificant.
From these conclusions, it seems highly likely that the brown surface seen on much of the stonework is an
organic residue of the linseed oil that is known to have been applied in the past. This residue seems to have caused
much darker discolouration and also served to reduce the permeability of the surface.
4.6 Salts
It has long been assumed that salts played some part in the deterioration of clunch. If this were the case, then it
would be expected that a decayed section of stone would contain salts which (due to the movement of salts with
moisture and the fact that most salt activity occurs within 20mm of the stone surface) might be expected to show a
gradient from surface to the interior of the stone. A core sample (40mm deep, 20mm diameter) was therefore taken
from Bay E04 (see Fig 8); this was divided into three sections and each section analysed (using BS standard for salt
analysis) by Dr. Mike Schwar of Cliveden Conservation.
The full analysis showed the following:
Location Front
(10mm
Middle (10- Back (30 –
30mm zone) 40mm
zone)
zone)
Nitrate, NO3: 0.002% 0.01% 0.02%
Sulphate, SO4: 0.5% 0.3% 0.4%
Fig 8: salt analysis from Bay E04
There was generally therefore a very low level of salts (with chlorides all less than 0.01%) and no clear gradient
(particularly sulphates) through the sample. The change in nitrate level was not thought to be statistically significant.
4.7 Cross sections of core samples
In order to try and understand further the way in which laminations develop, a further two core samples were
taken from Bay E04. These were mounted in resin at the Interface Analysis Centre at University of Bristol and were
examined using SEM/EDX. Scanning Electron Microscope (SEM) produced higher magnification images at
intervals along the prepared cross sections and Energy Dispersive X Ray analysis (EDX) produced information on
the elemental composition and distribution at various regions on the cross section.
One of the core samples came away in two sections – the main bulk and the lamination of the front face - but
the pieces fitted together well and so were then set together into resin. A x1.8 photograph of the cross section of this
sample (see Fig 9) is very instructive. The break between the front lamination and the main core can be seen as a
white line running across the section but there are also multiple minor laminations in the stone behind. Perhaps more
significantly, the area of the main bulk has a light colour and the dark area towards the front face shows that there
are some differences. NB the slight darkening around the edge of the whole sample is due to penetration of the
organic solvent for the resin in which the samples were set.
Fig 9: Cross section through core sample showing lamination near face

Energy dispersive X-ray analysis (EDX) targets an area about 250µm in diameter and records the amount of
each element present. The results of tests on the core sample not mounted in resin showed average content as laid out
in Fig 10:
Layer 1 (outer Layer 2 (Secondary Layer 3
lamination) lamination) (inside)
C 2.9 1.7 2.5
O 42.7 49.1 35.2
Na 0.3 0.3 0.2
Mg 0.8 0.6 0.7
Al 2.3 1.4 1.9
Si 10.0 6.6 8.7
P 1.4 0.9 1.3
S 0.2 1.7 1.3
Cl
0.1 0.1
K 0.3 0.1 0.2
Ca 38.1 36.8 47.1
Fig 10: Table showing average % content of elements at various depths of core sample
The areas affected by pore blocking have higher oxygen and lower calcium content. Without a greater number
of absolute measurements, it is not possible to make any firm conclusions. However it reinforces the idea that some
sort of change takes place near the surface which leads to the laminations seen in many of the clunch stones at
Woburn.
5. Common established repair methods
On many buildings constructed of clunch, evidence survives of the common practice of stone being protected
by limewash. On elevation E22 of the North Court of Woburn Abbey, traces of limewash were found and there is
archive evince of the use of milk and water being applied. As was seen in the analysis of surface microbiological
growth, there is also evidence of residues of linseed oil and this is thought to have had widespread use as a surface
preservative for clunch.
The 19 th century restoration of many of England’s churches was well intentioned but, although clunch may have
deteriorated due to long neglect, the methods of repointing, repair and rendering with cementitious mortars generally
led to even faster decay. The replacement of clunch with other (generally harder) stones also led to accelerated decay
of the adjacent softer clunch.
In recent years, the repair of clunch has been largely based on some well-established techniques of limited
replacement, cutting back decayed surfaces, lime mortar pointing, repair and rendering, and sheltercoating or
limewashing. The success of any or all of these treatments will depend on many factors not least the skills of the
conservators. However inspection of a number of buildings allows some limited conclusions to be drawn.
- Although clunch is still available, stone from the more durable seam is hard to obtain so the replacement of
stone has usually been restricted to structurally significant areas such as window mullions. In general these
have survived well.
- Cutting back the decayed surface (either of individual stones or of whole facades) is effective at removing
the laminated surface and providing an appearance that reflects the original architectural intention.
However, it changes the relative dimensions of mouldings and other detail. The technique of rubbing back
the surface also appears to fill the microporous surface with fine stone dust; this is particularly apparent if a
wet ‘spinning’ technique is used.
- Once the surface has been tooled back, there is also good evidence that the decay mechanisms become reestablished
and the laminations start again.
- Render repairs of decayed surfaces have had mixed results with exposed feather edges failing and
detachment of areas of render on stonework that is susceptible to regular wetting and drying.
- Patch render repairs of individual stones were not as successful as those where the whole face of a stone
had been repaired.
- Repointing is generally effective although, on a practical level, the high permeability of the stone means
that control of the drying of the mortars is difficult.
- Sheltercoats have been used to protect the stone and were always intended as part of a maintenance regime;
they will last only a few years in normal exposure.
- Limewash has been used but on smooth surfaces with a closed pore structure, it has a tendency to detach
within a short time particularly on exposed areas.

- In one location, many coats of limewater were applied after cleaning and levelling the stone. It created an
effect similar to a clear varnish to the stone face.
6. Trials carried out at Woburn Abbey in 1985
Practical Building Conservation Volume 1: Stone Masonry by John Ashurst was published by English Heritage
in 1988. Consolidation of Clunch in the South Court of Woburn Abbey was included as one of the case studies.
‘During August 1985 a series of comparative consolidation treatments were undertaken on the clunch of the
south stable block at Woburn Abbey. The stone had weathered and deteriorated, creating extensive areas of flaking.
It had been decided that dressing back was the most satisfactory treatment for this deterioration. As this is a process
which can seldom be repeated, requires all stones on a façade to be treated and is relatively labour intensive, it was
decided that the Research and Technical Advisory Service of HBMCE should investigate alternative procedures’.
The series of treatments were undertaken on the west elevation of the south stable block. The treatments were
chosen as being representative of the two sides of the debate over the suitability of traditional (and reversible) lime
wash versus the modern and irreversible alkoxy silane treatments. Linseed oil was put forward by the house restorer
at the time Cecil Rhodes and microcrystalline wax was chosen as a simple hydrophobic coating. One half of this
elevation had been dressed back and the other had not. The same treatments were applied to both types of surface
and the central pedimented bay was not treated.
A control panel was left within the dressed-back zone and the as-found zone. The weathered zone was first
brushed down with phosphor bronze brushes, treated with quaternary ammonium biocide and then brushed down
again. The treatment areas were to the internal west facing elevation of the South Court (Bays E33 - 35) as in Fig 11.
Fig 11: Diagrammatic representation of the trial treatments carried out in 1985
An appraisal of the treatment carried out six weeks after application (and included in Practical Building
Conservation) concluded that:
- Brethane (Areas 1 and 10) – this was found to have some darkening but was considered insignificant when
compared to the weathered stone adjacent. The surface felt very ‘tight’ and there was no dusting. It was felt
that the simple dressing and chamfering of the skin was acceptable when coupled with consolidant in this
way.
- Lime treatment (undressed surface) (Areas 3 and 4) – The techniques of grouting and filleting were
considered very satisfactory in terms of appearance and that much of the original surface was retained. It
provided a very sound surface compared to the untreated control.
- Lime treatment (dressed surface) (Areas 5 and 6) – These were significantly darker than the control –
probably due to saturation during treatment. Only a small amount of dusting was found on the surface.
- Boiled linseed oil (Area 7) – This had darkened but the surface was exhibiting very little dusting.
- Microcrystalline wax (Area 8) – Some accentuation of the redressing marks was recorded and some
blotchiness where the wax had been applied full strength. Generally better where the wax was dissolved in
white spirit.
- Control area (Area 9) – The surface was found to dust readily.
In 1985, it was concluded that: ‘the most satisfactory treatment for weathered clunch surface was considered to
be limewatering, filleting, grouting and mortar repair to the as-found surface followed by the application of shelter
coat’.

It was intended that appraisals should be carried out each year but the next recorded appraisal was by Malcolm
Starr of English Heritage in 2006. He wrote:
‘The response to this (original) recommendation was that the technique relied too much on maintenance, and
that this approach of what might be termed “total” conservation tended to change the original ashlar character and
precise architectural detailing of classical buildings. It was never adopted at Woburn.
All the chemical treatments had discoloured the faces, although material that had been exposed after brushing
off looked more natural in colour. The discolourations, and unknown long term effects, were felt to militate against
chemical treatments, in addition to the cost of some of them. Where original ashlar faces had been retained on the
lime method panel, large-scale delamination had continued. Small-scale delamination had continued on most of the
brushed back areas, and the mortar fillets had begun to detach in place’.
The paper then goes on to speculate that: ‘much of this deterioration might not have occurred if the work had
been maintained. The time span suggests that two further shelter coats might have been needed during the 23 years
since the trials were carried out’.
7. Review of trial areas in 2010
It is unusual to have the opportunity to review a well-documented treatment but an investigation was carried out
in 2010 to identify the effect of each of the treatment. The investigation used field techniques of observation backed
up by permeability measurement using a Karsten tube.
7.1 Observations
Area
no:
Treatment in 1985 Observation
1 All loose and flaking material was
removed. Edges of all flakes were
cut back to sound material using a
chisel to form a neat splayed
edge. Panel was thoroughly
2
brushed down prior to treatment.
Control – no treatment (except
biocide)
3 Lime poultice (lime putty applied
to thickness of 15mm, sheeted in
plastic and removed after two
days), lime water (40 No
applications) and lime mortar
repairs and grouting (stone
consolidated without redressing or
removing surface; edges were
supported with lime mortar fillets
and voids grouted).
4 Lime poultice, limewater, mortar
repairs and grouting (all as area 3)
plus lime casein shelter coat
5 Lime poultice, limewater, mortar
repairs and grouting (all as area 3)
plus lime casein shelter coat.
6 Lime poultice, limewater, mortar
repairs (all as area 3)
7 Boiled linseed oil applied by
brush (similar to the treatment
applied at that time to all new
stonework at Woburn)
8 Microcrystalline wax dissolved in
white spirit and applied by brush.
Then well rubbed in with a soft
cloth and surplus removed.
In the 25 years since application, there has been no further
decay except under a x10 magnifier, it is possible to see very
small flakes beginning to develop at low level. Because of the
presence of a lead catalyst in the Brethane, there has been no
microbiological growth on the surface.
Lamination and flaking has continued. Brown surface of
biological growth indicates biocide had limited effect.
Although some laminations remain intact, almost all of the
mortar fillets have detached or decayed. All laminations sound
hollow indicating that grouting has not arrested the separation.
Colour of surface slightly lighter than control.
Apart from tiny traces in protected areas, there is almost no
evidence of any shelter coat remaining. Although some
laminations remain intact, almost all of the mortar fillets have
detached or decayed. All laminations sound hollow indicating
that grouting has not arrested the separation. Colour of surface
slightly lighter than control
When compared to the control area, there is no obvious
difference in appearance or extent of decay. Only very small
traces of shelter coat remain in protected areas.
When compared to the control area, there is no obvious
difference in appearance or extent of decay
This bay has continued to decay at a faster rate than either the
control area (area 9) or the area treated with wax (area 8).
There is some surface discolouration when compared with
adjacent areas.
There is no noticeable discolouration of this area although it
still appears to retain a slight element of water repellence (see
‘Permeability’ section below). There appears to be less ongoing
decay than in the control area.
9 Control – no treatment Lamination (albeit localised) continues to develop.

10 Brethane There has been no deterioration of the surface at upper level
where the stone remains clean but at lower level, the surface
has darkened when compared to the adjacent control area.
7.2 Karsten tube tests
Although conclusions are limited from the permeability tests, a number of trends are clearly shown:
- Permeability on the sections treated with Brethane showed no take up of water at all after 10 minutes; this
showed that the hydrophobic qualities of the consolidant are still intact.
- All measurements taken in Bay E35 (areas 1 - 4 not dressed back) show higher permeability compared to
those in Bay E33 (areas 5 – 10 dressed back).
- The control area (area 9) also has low permeability indicating that the dressing back serves to close the
surface and reduce permeability.
- The most permeable samples are all on areas where the original surface has been lost (areas 2, 3 and 4).
- The least permeable stone (apart from that treated with Brethane) is that treated with oil.
- There is no discernible difference in permeability between the areas treated with lime poultice, limewater
and shelter coat and the adjacent control area.
- The presence of lichen reduced the permeability of the surface.
7.3 Conclusions from the trials
The result of the detailed inspection of the trials showed that many of the assumptions of the decay mechanism
were correct. After 25 years, dressed back areas show signs that the decay mechanism has been re-established. Lime
treatment (limewater, mortar repairs and shelter coat) does not appear to mitigate the causes of deterioration but, in
consideration of its use elsewhere, it can provide short term stabilisation of the surface. If regularly repeated as part
of a maintenance programme, then this stabilisation could reduce the extent of decay. Of perhaps most interest is that
the areas treated with Brethane (that effectively prevent liquid moisture transfer into the stone) have been effective at
slowing down the cycle of decay. Re-treatment would be problematic not only because Brethane is no longer
available but it is not known how other treatments would interact with the hydrophobic surface. Other evidence
suggests that any method that slows down (rather than prevents) the absorption or expulsion of water (e.g. oily film,
possibly lichen, tooling the surface) will exacerbate the pore blocking effect and thus the cycle of decay will
continue. A roughened surface, either from tooling (not spinning) or natural decay of surface lamination, provides a
more permeable surface that is less susceptible to pore blocking.
8. Options for treatment of decayed clunch
In the light of the review of the trials carried out Woburn in 1985, the field and laboratory investigations into
current decay and observations of treatments to other clunch buildings, it is possible to set out the advantages and
disadvantages of the various treatment options.
Methodology Advantages Disadvantages
Consolidation Although depth of penetration is likely
to be limited, consolidation which
results in a completely hydrophobic
surface appears to be effective and the
stone surface appears stronger to the
touch
Repair with lime mortar Use of compatible repair mortars can
protect decayed sections and secure
laminations
More effective in areas which are more
protected from wetting/drying cycle
Surface treatment with
limewash
Surface treatment with
shelter coat
Provides homogenous aesthetic
appearance
Provides short term protection to stone
surface
Provides homogenous aesthetic
appearance for short period
Hydrophobic surface means that other
(water based) treatments cannot be
carried out
Re-treatment may be difficult and have
little benefit
Evidence suggests that mortars will tend
to detach within a few years so repair
should be seen as a maintenance activity
Matching mortar with the fine-grained
stone may produce a repair with a closed
surface. Evidence suggests that an open
textured permeable mortar is more likely
to be of benefit
No evidence that it slows down decay
mechanisms
Likely to require re-treatment within ten
years
No evidence of any beneficial effect on
decay mechanism
Porous aggregates that are used to make
shelter coat will have to be very finely

Surface treatment with
linseed oil
A traditional treatment but mainly for
reducing premature drying out of
mortar used in fixing
Surface treatment with Possible reduction in decay due to
wax
hydrophobic qualities
Removal of surface Removes existing layer of decay
Provides ‘as new’ appearance
If surface is left rough or laitance
removed from pores (for example by
using dilute acid), this may allow better
moisture transfer and reduce pore
blocking
Replacement of stone Reintroduces correct planes and
profiles of masonry
Reintroduces structural integrity
Surface left roughly tooled may
encourage moisture movement and
reduce pore blocking
9. Conclusions
ground to be able to fill the pores of
microporous stone which may anyway be
blocked
Causes discolouration
Evidence of slightly increased rate of
surface decay
In time, this may encourage certain types
of lichen growth
May cause some discolouration
Will break down with exposure
Affects relative dimensions of
architecture
Unsustainable as cannot be repeated
Technique of spinning assists in
establishing new ‘pore blocking’ cycle of
decay
Removes historic fabric
Durability of new stone may be no better
than that removed
Fine finishing of surface likely to
encourage early pore blocking and lead to
decay
The treatment of decay of stone must depend on many factors such as a sound analysis of the causes of decay
and an assessment of the most suitable materials for conservation. It is often the case that there is no opportunity to
review these treatments and, in most cases, the review will be superficial and subjective. In this programme of work
at Woburn Abbey, we were able to carry out a more complete assessment of the causes of decay through field
observation and laboratory analysis and to conclude that pore blocking is the most dominant contribution to decay.
We were able to show the value of surveying to establish how those areas exposed to more regular moisture
movement suffered from a greater degree of lamination. Through consideration of permeability measurements, it
was possible to establish that anything that closed up the pores of the stone (for example tooling off the surface and
particle hydrophobic treatments such as linseed oil) also led to more loss through lamination. On the other hand
natural decay of the surface to provide an open texture or interventions that protected the surface and yet allowed
moisture movement (for example limewash) would seem to reduce lamination. Complete hydrophobic treatment (for
example brethane and wax) also appeared to reduce the decay rate although there are significant doubts about
continuing maintenance and re-treatment.
It would appear then that the centuries of the tradition of ‘sheltercoating’ a stone with lime wash to protect it
from weather was well founded. It can only be speculation as to what the result would have been had the trials
carried out in 1985 been maintained by regular limewashing. For the moment, the advice for the repair of clunch
should be that it be kept clean and free of surface dust, grime and biogrowths and to undertake interventions that
maintain a permeability rather than allow an impermeability.
The study of the trials at Woburn Abbey has shown the benefit of being able to look in detail at a variety of
treatments and provide some objective field measurement which will help to back up a comprehensive visual record.
Although resources will often prevent such re-visits occurring, it can only be of benefit that what sometime seem
like instinctive decisions on treatment can be supported by sound evidence.
10. References
Ashurst, J., Rhodes, C. 1991. Course notes on ‘Durability and Weathering of Clunch’
BRE Stone testing Data sheet http://projects.bre.co.uk/ConDiv/stonelist/totternhoe.html
Ashurst, J. 1988. Practical Building Conservation: English Heritage Technical Handbook, Vol 1: Stone
Masonry. London: Gower Technical Press
Dimes, F. 1990. Conservation of Building and Decorative Stone Part 1. London: Butterworth Heinemann
Starr, M. 2006. Clunch - Case Studies. Paper presented at the Institute of Conservation Stone and Wall Painting
Group conference ‘Problem Stones’ at Tower of London 2006.

Marble statues and panels of San Petronio façade in Bologna. State of
conservation after 40 years from restoration
Susanna Bracci a , Monica Galeotti, Daniela Pinna b,*
a
Istituto per la Conservazione e la Valorizzazione dei Beni Culturali – CNR, Via Madonna del
Piano, 10, 50019 Sesto Fiorentino (FI), Italy
b
Opificio delle Pietre Dure, Ministero per i Beni e le Attività Culturali, viale Filippo Strozzi 1,
50<strong>12</strong>9, Firenze, Italy
* Contact author’s e-mail address: daniela.pinna@beniculturali.it
Abstract
In the 1970s the marble statues and panels on the façade of San Petronio Church in Bologna were
restored by Ottorino Nonfarmale. The cleaning and treatment operations were chosen after an
accurate scientific examination performed by Raffaella Rossi Manaresi and other conservation
scientists. The façade of San Petronio Church is going again to undertake a restoration and this
event has been the occasion for studying the conservation state of statues and panels after about 40
years from the past restoration. Some samples taken in the 1970s, before Nonfarmale’s intervention,
from panels and sculptures of the facade have been examined focusing on the existing patinas and
deposits. The results have been compared with those obtained from samples taken in 2010.
The final step of the restoration carried out in the 1970s consisted of the treatment of marble
surfaces with a mixture of an acrylic resin copolymer (Paraloid B72) and a partially prepolymerized
dimethylpolysiloxane (Dri Film104) in solvent, which gave both water-repellency and
consolidation to the stones. Thus a further aim of this study has been the evaluation of the treatment
after about 40 years of natural ageing.
Both past and recent samples have been analyzed with various techniques: stereoscopic
observations, microscopic examination of both thin- and cross-sections under visible light and UV
radiation, FTIR spectrometry on KBr micro pellets, micro FTIR on cross-sections, ESEM
connected to an EDS probe. Moreover petrographic analyses were performed on thin-sections.
The results allowed to assess the present state of conservation of marbles and to compare it with the
one before the past intervention. Moreover the penetration depth and distribution of the treatment as
well as its long-term water-repellency has been evaluated.
Introduction
The Gothic Church of San Petronio is dedicated to Petronio, the 5th century bishop and patron of
Bologna. Dominating the south side of the city's main square Piazza Maggiore, San Petronio
Church was begun in 1390 as a civic project and was completed in 1663 (1). The façade - 51 metres
in height and 60 metres in width - is only less than half covered in polychrome stones and
decorations. The rest, made of bare bricks that should have been covered with marble, remained
unfinished. Nevertheless, the finished half is of exquisite beauty with sculptures by Jacopo della
Quercia decorating the Porta Magna, the main doorway, between 1425 and 1438. They depict the
Madonna and Child, San Petronio, St. Ambrose and biblical scenes (2). The side portals were
elevated instead between 1518 and 1530. The lunette of the left portal is decorated with the
Resurrection carved by Alfonso Lombardi, while the right one shows the Deposition of Christ by
Amico Aspertini, the Virgin by Tribolo and St. John by Ercole Seccadenari (3).
The church underwent over time several restructuring and restoration interventions. The first
documented restoration on the stone façade was conducted by the restorer Ottorino Nonfarmale
from 1972 to 1979. At that time the facade was covered with a dark, compact, homogeneous crust,
detached only in some jutting out and exposed areas. In some sculpted parts of side portals, pieces
of carved stone had fallen down (fig.1) (4).

In view of the 350th anniversary of San Petronio’s completion in 2013, the “Fabbriceria” (council
of maintenance) worked out a special intervention for completing the restoration work that was
already carried out in many other parts of the church. The intervention consists in an extraordinary
restoration plan of pieces of art and architecture, to be carried out between 2010 and 2013. The
project has been named “Felsinae Thesaurus” (the treasure of Bologna), as this is how San
Petronio is defined in an inscription engraved on a memorial tablet present on the exterior wall of
the chapel consecrated to him in Archiginnasio street in Bologna: “Pone lapidem Felsinae
Thesaurus”. So, after about forty years since the last restoration, the façade of San Petronio Church
is going to be restored again. The conservation project started in September 2010 and included
preliminary scientific investigations, run by the “Opificio delle Pietre Dure” in collaboration with
the Institute for the Conservation and Valorization of Cultural Heritage - CNR. The scientific study
focuses on the three portals, but the conservation state of the whole façade was evaluated,
including the upper part made of bricks. OPD is nowadays in charge of the restoration of the
façade.
This paper deals with the study of some samples taken in the 1970s, before Nonfarmale’s
intervention, from panels and sculptures of the facade focusing on the existing patinas and
deposits. The results were compared with those obtained from samples taken in 2010.
Materials and methods
A large number of samples - 6 collected in the 1970s and 39 in 2010 - were examined, but only the
results referring to the most significant ones are discussed in this contribution, i.e. three samples of
the portals collected in the 1970s before the intervention, and three samples taken in 2010.
The following analyses were carried out on the samples:
- observation under stereo- and optical (visible and ultraviolet radiation) of thin (30–40 µm) and
cross-sections; the sections were obtained after including the samples in polyester resin;
- Fourier Transform Infra-red Spectrophotometry on KBr micropellets using a IR
CONTINUUM TM microscope; micro FTIR in reflection mode was performed on the surfaces
of some samples. Micro ATR was carried out on some thin-sections to detect and better
localize organic compounds on selected layers;
- examination of thin and cross sections with an energy dispersive X-ray spectrometer (EDS)
connected to the scanning electron microscope (ESEM Quanta200 - Environmental Scanning
Electron Microscopy) to study the interactions between patinas and stones, as well as the
conservation state of the latter. The instrument allows non-conducting specimens to be imaged
and analysed without coating them since it is possible to acquire either SEM images or EDS
analyses in low vacuum (0.1 torr).
Moreover, thin sections were petrographically studied with a mineralogy microscope (Zeiss
Axioscope A1).
Results
Diagnostic investigation preliminary to the restoration of the 1970s
At the time of San Petronio façade restoration, the studies on alteration and conservation of stone
were very scarce in Italy and abroad. Cesare Gnudi took the opportunity of San Petronio
restoration to promote the research in this field founding in Bologna the Centro per la
Conservazione delle Sculture all'aperto. Thus, in that occasion, historical researches and scientific
investigations were carried out before the intervention in order to identify the materials of the
façade, to define their state of conservation and the causes of alteration, and to elaborate a correct
method of restoration. Dr. Raffaella Rossi Manaresi run and coordinated the work with the
collaboration of several research institutes. Hereinafter the main results, obtained with the
techniques available at that time, are summed up.
The analyses carried out on samples from panels and sculptures of the central portal showed that
the marble was generally well preserved and still compact, and a dark superficial crust was

distinctly overlying on it (fig. 2) (4). The marble was of good quality, compact, low-porous.
However, the state of preservation could hardly be explained only with the stone’s good quality.
According to the researchers a hydrophobic material was probably applied preventing the
penetration of water and thus avoiding alteration processes took place. Actually they detected a
brown film on the stone which penetrated micro-fissures at a depth of few microns, suggesting the
hypothesis of an ancient, perhaps original treatment (6). The brown film was covered by a dark
crust, made of two layers in some samples. The crust contained black particles, traces of ferric
oxides, a little amount of calcite and quartz and a great amount of gypsum (in some cases 43 % of
gypsum, 2.5 % of calcite and traces of magnesium, sodium and potassium salts) (7). On the other
hand only calcite and no gypsum were detected under the crust. Such observations led the
researchers to hypothesize that the crust couldn’t have caused by the deterioration of the stone, and
to suggest that it should have been linked to treatments (6).
Fig. 1 Panel depicting “The Expulsion from the Garden of Eden” before the 1970s restoration (5). The dark crust homogeneously
covering the stone was fallen off in some of the most exposed areas.
Fig. 2 Cross-sections of samples from central portal. The division between the dark crust and the underlying stone is clearly distinct
(5).
The examination of the brown film overlying the carved stone resulted in a complex issue.
Beeswax, a natural resin such as sandarac or colophony, substances identified as belonging to rue
leaves and, only in one sample, calcium oxalates were detected. There was also a high amount of
triglycerides, higher than that of standard beeswax. Since triglycerides are the fundamental
constituents of fats and oils, they might be linked to those substances (8). Moreover proteins in
samples from two panels of the main portal were detected (8).
The use of both sandarac and rue oil was reported in documents attesting a payment to Jacopo
della Quercia on July 15 th 1428 per vernixe e per ollio de ruda da unzere le colonne rosse (for
varnish and rue oil to spread on the red columns). An analogous document for the sculptures was
unfortunately not found. Moreover it is worth mentioning that wax-based treatments applied to
protect the outdoor sculptures are documented since ancient times. Vitruvius for example
suggested to mix wax and oil, and also wax and resin (6).

Samples from the side portals were also examined (9). They were covered with a homogeneous
very compact dark crust. Some details of the sculptures were detached or damaged. The crust had a
composition very similar to the one of the central portal, but it contained also casein. As it hadn’t
the characteristics of an alteration crust, it was probably caused by a treatment of the sculpted
surfaces. The presence of iron-based pigments and of casein, which could acts as binding medium,
confirmed this latter hypothesis.
The brown organic film in contact with the stone and the beeswax, found on the central portal,
were not detected.
All the three portals had surely undergone many treatments over the centuries (10). The analyses
showed at least two coatings. One is the brown film containing beeswax and detected on the main
portal only. The general treatment (over the brown film) was hypothesized to be coloured (redorange)
and made of a mixture containing gypsum, or calcium hydroxide transformed into
carbonate and then into gypsum by atmospheric action; casein was used as binding medium (4).
Ceresin (a synthetic wax which came into use in the nineteenth century), found on samples from
the statue of Madonna (main portal), could explain the presence of hydrocarbons detected on all
the portals. It could have been applied perhaps when the last retouching had been done, even just
to obtain an aesthetic effect (4).
Oil, rarely identified in samples, could have been the binding medium of a coloured mixture used
to “retouch” the previous “painting”. This hypothesis was based on the microscopic observations
which showed two distinct layers corresponding to two successive treatments.
Thanks to historical studies and investigations, and to a comparison between some photographs of
the same panels taken on different dates, it was supposed that the first casein-based treatment was
applied in the period between the two World Wars while the retouching was done between 1950
and 1974 (4).
The aforementioned conclusions made it possible to elaborate the appropriate restoration methods.
The main goals were: the removal of the black crust, the consolidation of the materials in the most
degraded areas, the protection of the entire surface with a water-repellent product.
Cleaning operations were then performed using poultices of attapulgite with water, and poultices
of carboxymethyl-cellulose containing basic salts, EDTA, a surfactant (Desogen), and
triethanolamine. The consolidating and water-repellent treatment was obtained by mixing Paraloid
B72 (copolymer methyl acrylate-ethyl methacrylate, 30% w/w in nitre diluent), a silicon-based
compound (Dri Film 104, 70% w/w in white spirit), 1,1,1-trichloroethane and acetone in ratio
15/5/40/40 v/v, respectively (10).
Diagnostic investigations carried out in 2010-2011
As mentioned, the restoration project “Faelsinae Thesaurus” included also preliminary scientific
investigations which started in September 2010. The statues of the main portal are made of
Candoglia marble, the ones of the side portals and the panels of Carrara marble, the lintels are
composed by Oolitic Limestone from San Vigilio (11).
A number of samples, both collected in the 1970s and in 2010, has been examined for this study,
but only the most significant ones are mentioned in this article. The re-examination of the 1970
samples and their comparison with the ones taken in 2010 were aimed at checking the past results
and deepening their study with the aid of other techniques in particular ESM/EDS and microFTIR,
able to provide further information on the composition of the single layers. Moreover the samples’
comparison revealed either the effect of the restoration on stone surfaces or the past and present
state of conservation.
Fig.3 shows the cross section of a fragment taken in the1970s from the back of the basement of St.
Ambrogio statue (main portal). In the visible light image a patina, composed by two layers, covers
the marble. The layer in contact with marble is 30-60 μm thick and brownish coloured, while the
other one overlying it, is discontinuous, 40-50 μm thick, light-grey coloured, and externally darker
with black airborne particles. Both layers contain red-orange particles (Fe silicates and Fe oxides).

UV light images (not shown) reveal a whitish fluorescence in some areas of the patina and inside
the grains of marble. Examinations of the thin section show that the stone is affected by inter and
intra-crystalline micro-fractures and that the layer in contact with the stone penetrates marble’s
micro-fractures. FTIR and EDS analyses enabled to differentiate materials of the external layer of
the patina from the ones of the inner layer. Amide I and amide II bands, featuring protein
substances, and phosphorus were detected on a brown material, perhaps belonging to the layer in
contact with the stone. Moreover, oxalates were specifically identified in the external layer of the
patina. Phosphorous and proteins may be associated to the treatment with casein used as binding
medium, as reported in the literature (4) (9).
The cross-section of a fragment collected from the same statue in 2010 is shown in fig.4. Despite
this sample comes from a different area, a comparison between the two samples can still provide
insights into the changes occurred since the previous investigation. The marble, again microfractured,
is covered by a compact, light brown patina 30 μm thick (labelled 1 in fig.4). It is not
continuous even though, when present, it looks adhering very well to the stone and partly
penetrating the inter grain fissures (fig.5). Combining ESEM-EDS and FTIR results, gypsum,
allumino-silicates, a small amount of phosphorus and oxalates were detected in the patina. The noncolored
layer over the patina (labelled 2 in fig.4) contains instead gypsum, siloxane and acrylic
resin. Acrylic resin was identified by IR bands at 1730, 1480 and 1380 cm -1 , and siloxane by IR
bands at <strong>12</strong>74, 850 and 780 cm -1 (<strong>12</strong>). Other bands, which according to Favaro et al. (<strong>12</strong>) are useful
to evaluate the degree of degradation of the polymer mixture, are hidden by gypsum and silicates
vibrations, and analyses of the extracts are still in progress. Nevertheless, the well detectable <strong>12</strong>74
cm -1 band, assigned to the Si-C stretching of the group Si-CH3, indicates that un-reacted alkoxy
groups are still present in the polymeric mixture.
Fig.3 Cross-section of a sample taken in the 1970s from Fig.4 Cross-section of a sample taken in 2010 from the statue
the statue of St. Ambrogio (main portal). of St. Ambrogio. 1- light brown patina (30 μm), 2-
non-colored layer with gypsum, siloxane and acrylic resin
(some glass spheres are visible on the right), 3- top black
layer.
The top black layer contains mainly round-shaped carbon particles, gypsum and silicate, likely a
deposit of airborne material. Furthermore glass spheres, 50 μm in diameter, are present under the
black external layer (fig.4). They are an evidence of the mechanical cleaning carried out with airabrasive
during the 1970s restoration.
Thus the re-examination of the 1970s sample and its comparison with the one taken in 2010 show
the two-layered patina occurring on stone before the restoration (confirming the 1970s
investigations). The new examination brought to a more precise definition of its origin and
composition. It was indeed the result of two intentionally applied treatments based on organic
substances: a protein-based treatment for the first layer and an oil-based one for the second layer (as
it can be inferred by the oxalates). Since the first layer penetrates the marble micro-fractures, it is
unlikely that it was originally applied on the sound marble. Moreover the analyses show that it was
partially preserved during the cleaning carried out in the 1970s, probably because it had proven to

e effective against the marble’s decay. The glass spheres and the mixture siloxane-acrylic resin are
mostly embedded in it or laid on its top, confirming that the removal of the superficial material was
limited to the dark external layer. The current outer black layer was therefore formed over the last
decades as a result of dry deposition while the marble underneath was not affected by a further
weathering.
On the statue of the Madonna on the main portal, the marble is covered by a mono-layered greybrownish
coloured patina (40-50 μm), containing orange and black particles, visible both in the
1970s sample and in the one of the 2010. Combining ESEM-EDS and FTIR results, gypsum,
allumino-silicates, Fe oxides, phosphorus and oxalates were detected in the patina. The composition
is similar to that of the previously discussed samples. The 2010 sample contained also siloxane and
acrylic resin. A dark powdered atmospheric deposit with inhomogeneous thickness is visible on the
patina. In both samples the marble is highly weathered but not affected by gypsum formation in
between the grains. While gypsum occurs mainly inside the black deposit, silicon is instead layered
into the patina as siloxane (see EDS maps of Si, Al and S in fig.6) and also dispersed in the deposit
as silicate particles. Fig.7 shows an ESEM picture of the patina’s surface, where some flat areas
composed of acrylic resin and siloxane are visible. They are covered by acicular and tabular (to a
less extent) gypsum crystals and alumino-silicates aggregates. The coating covering the marble is
still quite uniform and its appearance is similar to a newly treated surface (<strong>12</strong>). It is worth
mentioning that the whole depth of the superficial layer (patina + deposit) in both samples, is less
than 150 μm and that the morphology of the cross-section is quite the same even after many years
and an extensive conservation work. It seems that the marble has not undergone a further decay
since the 1970s.
Fig.5 Back- scattered electron micrograph of a detail of Fig.6 Back- scattered electron micrograph of the cross-section of
cross-section in fig.4. The compact, well adhering patina a sample from the Madonna statue (main portal). EDS maps of Si
is visible. The EDS spectrum shows Ca, Si, S, Fe, Al, (top right), Al and S (bottom left and right).
Mg, K and P.
Carrara marble panels of the side portals are much more jutting out than the statues which are
protected by the lunettes. Two samples from the panels of the right portal have been compared. The
investigations carried out in the 1970s on samples’ cross-sections showed two layers (more rarely
three) over the stone. This result is observable in the cross-sections of a sample taken in the 1970s
from the panel depicting “the Coronation of the Virgin” (fig.8) and of a sample taken in 2010 from
the panel depicting “The Flagellation of Christ” (fig.9). In both of them, over the micro-fractured

stone, a two-layered patina is visible, where the layer in contact with marble is thin (10-20 μm),
compact, brown coloured and partially penetrating the marble micro-fractures. Combining ESEM-
EDS and FTIR results, gypsum, allumino-silicates, Fe oxides, traces of oxalates and a natural wax
(identified by the bands at 1730 cm -1 , ~1465 cm -1 and ~1053 cm -1 ) were detected in the layer in
contact with marble. The 2010 sample contained also siloxane and acrylic resin. Micro-FTIR with
ATR on thin-section of the older sample reveals, along with silicates signals, bands at ~1540 and
1504 cm -1 assigned to carboxylates. These last results are likely an evidence of altered oil or wax.
The 2010 sample also contained lead and phosphorous, both detected by EDS (fig. 10). Over the
layer in contact with stone, a thicker (from 70 to 100 μm in both samples) grey coloured one is
visible. Despite the fact the elemental composition (Ca, S, Si, Al, Fe) of both layers is the same, the
external one contains a larger number of orange, white and black particles.
The grey layer of the 1970s sample goes under the stone fragment just like if it had entered fractures
of the stone, favouring the detachment of little scales. It is covered by a black deposit.
The EDS map of the 2010 sample shows that gypsum is mainly located on the outmost surface. It is
probably connected to dry deposition along with black particles also observed. The results obtained
on samples from the side portals do not fully match those of the1970s research reporting a monolayered
patina and in particular the lack of the beeswax protective coating detected instead on the
main portal. On the contrary our examinations of old and recent samples from the side portals
clearly show a thin, organic-based treatment which preserved the underlying marble over time.
Lead detected in the sample from “The Flagellation of Christ” could be related to a remnant of a
“scialbatura” layer.
The cleaning carried out in the 1970s did not intentionally affect the layer in close contact with
marble, whose preservative efficiency seems to be indeed strengthened by the resin mixture used by
Nonfarmale. As a matter of fact, siloxane and acrylic resin were extensively detected by FTIR on
almost all the 2010 samples, not only on marble but also on Verona stone slabs, which are not
sheltered. The EDS distribution maps of silicon in cross-sections show that it is located in the first
layer on the stone where, not being associated with aluminium and other elements of silicate
compounds, it is therefore related to siloxane. In some cases, depending on the preservation state of
the underneath stone, silicon penetration in the marble is around 0.8 mm. On the other hand, sulphur
inside the marble and at the interface with patina is rarely detected, showing that sulphation
occurred only at a minor extent. In situ reflectance FTIR spectroscopy (data not shown) allowed to
much enlarge the number of investigated areas in a non-invasive way and to confirm the widespread
distribution of Dri Film 104 and Paraloid B72. Tests with the sponge contact method were
performed (data not shown) to evaluate the long-term performance of siloxane (13). The results
showed that the values of treated surfaces are close to the ones of sound marble, demonstrating the
treatment is still performing in terms of water-repellency.
Fig.7 ESEM micrograph of the surface of a sample from the Madonna statue. The polymer coating and acicular and tabular gypsum
crystals are shown.

Fig.8 Cross-section of a sample taken in the 1970s from the panel depicting “the Coronation of the Virgin” (right portal).
Fig.9 Cross-section of a sample taken in 2010 from the panel depicting “The Flagellation of Christ” (right portal).
Fig.10 Back- scattered electron micrograph of a detail of the cross-section in fig.9. EDS spectrum of a spot where Pb was detected.

Conclusions
The stone surfaces of San Petronio façade, treated about 40 years ago with a mixture of an acrylic
polymer and siloxane, were investigated and the results were compared with those ones related to
the examination carried out before the 1970s restoration. Moreover, samples taken at that time were
re-examined, and stone’s changes as well as treatment performance over time were assessed. A thin,
organic-based layer in close contact with marble was detected on samples before and after the
1970s intervention, both from the central and the side portals. A second thicker layer was placed
over it. The layers contained, among other compounds, oxalates, i.e. the end-products of chemical
transformation over time of more complex organic substances applied to protect the stone (14).
Thus they were related to an intentional treatment with organic substances mixed with red-orange
particles (Fe silicates and Fe oxides). The thin organic-based layer in close contact with marble,
detected before Nonfarmale restoration, resulted to be partially left after that.
Dark airborne deposits affected the outmost surfaces while sulphation phenomena seemed to be
very limited.
Siloxane and acrylic resin impregnated the pre-existing treatments and penetrated the microfractures
of marble, thus performing a further protective and consolidating action, avoiding in this
way more massive alteration of the stone. The components of the mixture applied in the 1970s are
indeed still present in a fairly high content and, as far as the analytical results achieved so far allow
to infer, are not affected by deterioration or strong molecular changes. Studies on monuments
treated in the 1980s with Paraloid B72-Dri Film 104 mixture (15, 16, 17, 18) provided evidences of
loss of water-repellency of treated stone, strong reduction of the content of siloxane, formation of
low molecular weight compounds originating from alteration of the acrylic component (crosslinking
and chain scission). These results were confirmed by the examination of marble specimens
treated with Paraloid B72-Dri Film 104 mixture and artificially aged through photo-oxidation (15).
On the other hand R. Rossi Manaresi herself studied the effect of acid fog (using 0.02 M H2SO4)
and UV radiation on limestone samples treated with Paraloid B72-Dri Film 104 mixture and then
artificially aged (19). While UV radiation did not alter the morphology of the treated surfaces,
significant changes appeared instead after acid fog exposure resulting in a three-dimensional
network of the treatment, thicker than non-aged one, with small cells. This change was observed
though only on surfaces directly exposed to acid fog and the authors ascribed the result to the fact
that in situ polymerization of the silicone is triggered by the weathering itself.
Actually the results obtained examining San Petronio façade are partially in contrast with the
mentioned literature. This can be tentatively explained by noticing that the photo-oxidative
deterioration of the mixture is strongly affected by the application method. Tests reported by Favaro
et al. (15) showed that the mixture’s degradation is enhanced whenever it is applied as thin film on
marble. On the contrary, polymer modifications resulted limited when the mixture is applied as
thick coating, as it is likely the case of San Petronio where the bulk of the thick polymer layer is
probably not affected by oxygen penetration. Moreover, most sculpted surfaces are nowadays
uniformly covered by atmospheric deposits which could have provided a protection from the direct
exposure of acid fog. Finally, a synergic action of both the polymer mixture and the old organic
treatment could also be invoked to explain the long lasting water-repellency of the surfaces.
Acknowledgements
The authors wish to thank Monsignor Oreste Leonardi, Primicerius of San Petronio Church, for
permission to collect samples. The authors acknowledge the support of this work by Maria Cristina
Improta Director of the Stone Restoration Department at Opificio delle Pietre Dure, by Roberto
Terra and Guido Cavina Directors of the restoration of San Petronio Church, and by Gian Carlo
Grillini who carried out the petrographic examinations of stones.
References

1) M. Fanti, 1983. La basilica di San Petronio nella storia religiosa e civile della città – Genesi,
vita e significato del monumento, In: La Basilica di San Petronio in Bologna, Cassa di Risparmio di
Bologna, Bologna, v. I: p. 9-27.
2) R. Grandi, 1983. Cantiere e maestranze agli inizi della scultura petroniana, In: La Basilica di
San Petronio in Bologna, Cassa di Risparmio di Bologna, Bologna, v. I: p. <strong>12</strong>5-131.
3) M. V. Brugnoli, 1984. Le porte minori, In: La Basilica di San Petronio in Bologna, Silvana
Editoriale, Bologna, v. II: p. 61-82.
4) C. Gnudi, R. Rossi Manaresi, O. Nonfarmale, 1979. Report on the conservation of the facade of
S. Petronio, Centro per la conservazione delle sculture all’aperto, Edizioni Alfa, Bologna.
(5) R. Rossi Manaresi (editor), 1981. Jacopo della Quercia e la facciata di San Petronio a Bologna.
Contributo allo studio della decorazione e notizie sul restauro. Edizioni Alfa, Bologna.
(6) R. Pellizzer, R. Rossi Manaresi, 1970. Sullo stato di conservazione del portale maggiore della
chiesa di S. Petronio a Bologna, Atti Acc. Fisiocritici Siena, Serie XIV, v. II, p. 269-283.
(7) M. Tabasso Laurenzi, 1971. Studi e osservazioni sullo stato di conservazione del portale
centrale di San Petronio a Bologna, Proc. of the Int. <strong>Congress</strong> La conservazione delle sculture
all’aperto, Bologna 23-26 ottobre 1969, Ente bolognese manifestazioni artistiche, p. 117-132.
(8) S. Barcellona, M. Tabasso Laurenzi, 1971. Studio del materiale organico presente sulla
superficie del portale centrale di S. Petronio a Bologna, Internal report, June 1971, ICR, Roma.
(9) R. Rossi Manaresi, 1973. Sullo stato di conservazione dei portali laterali della Chiesa di S.
Petronio a Bologna, In: Problemi di conservazione, ICR, Roma, Editrice compositori, Bologna, p.
395-402.
(10) C. Gnudi, O. Nonfarmale, R. Rossi Manaresi, 1981. Il restauro della facciata di San Petronio,
In: A. M. Giusti (Ed.), Atti del convegno sul restauro delle opere d’arte – Firenze Novembre 2-7,
1976, Edizioni Polistampa, Firenze, p. 213-220.
(11) G. C. Grillini, 2011. Facciata della Chiesa di San Petronio. Identificazione dei litotipi.
Technical report, 2011, (not published).
(<strong>12</strong>) A.E. Charola, A. Tucci, and R.J. Koestler, 1986. On the reversibility of treatments with
acrylic/silicone resin mixtures, JAIC, Volume 25, Number 2, Article 3, p. 83 – 92.
(13) D. Vandevoorde, M. Pamplona, O. Schalm, Y. Vanhellemont, V. Cnudde, E. Verhaeven. 2009.
Contact sponge method: performance of a promising tool for measuring the initial water
absorption. Journal of Cultural Heritage,10(1), p. 41–7.
(14) G. Alessandrini, 2005. Patine sui materiali lapidei, In: P. Tiano, C. Pardini (Eds.), Le patine –
Genesi, significato, conservazione, Nardini Editore, Kermes quaderni, Firenze, p. 15-28.
(15) M. Favaro, R. Mendichi, F. Ossola, S. Simon, P.A. Vigato, 2007. Evaluation of polymers for
conservation treatments of outdoor exposed stone monuments. Part II: Photo-oxidative and saltinduced
weathering of acrylic-silicone mixture. Polymer Degradation and Stability, 92 p. 335-351.
(16) S. Haake, M. Favaro, S. Simon. 2004. The Bologna Cocktail and evaluation of past
consolidation treatments. In: Proceedings of the X Int. <strong>Congress</strong> on Deterioration and Conservation
of Stone. Stockholm, Sweden, ICOMOS, p. 423-430.
(17) M. Favaro, S. Simon, C. Menichelli, V. Fassina, P. A. Vigato. 2005. The four virtues of the
Porta della Carta, Ducal Palace, Venice. Assessment of the state of preservation and re-evaluation
of the 1979 restoration. Studies in Conservation, 50(2), p.109-<strong>12</strong>7.
(18) L. Appolonia, F. Bevilacqua, C. Di Francesco, D. Pinna. 2000. Verifica dello stato di conservazione di
monumenti ferraresi dopo alcuni decenni dagli interventi di restauro. In: Proceedings of the <strong>Congress</strong> “La
prova del tempo”, Bressanone, p. 163-171.
(19) A. Tucci, R.J. Koestler, A. Charola, R. Rossi Manaresi, 1985. The influence of acid rain and
UV radiation on the ageing of acrylic and silicone resin. In: Proceedings of the 5 th Int. <strong>Congress</strong> on
Deterioration and Conservation of Stone. Lausanne, Switzerland, Press Polytechniques Romandes,
p. 891-898.

OUTSIDE THE CANON: A REVIEW OF UNIQUE APPROACHES TO STONE
CONSERVATION
Justine Posluszny Bello, 1 Daniel Lane, 1 Mark Rabinowitz, 1 Joseph Sembrat 1 ,
1 Conservation Solutions, Inc.
Abstract
For the best of reasons, conservators tend to rely on tried and true means and methods
when treating stone at the expense of innovation. Failures with previous inventive
materials and applications have made us cautious about trying new approaches. As a
result, most practitioners rely on a relatively narrow range of proven treatments.
The authors will present examples of new and not-so-new treatments they have
performed using materials, means and methods that are outside of the generally accepted
canon of cleaning, biocidal treatments, structural repairs, surface treatments, and
material replications for outdoor stone monuments and sculptures. Some treatments
have been developed recently while others have been in service for decades yet remain
obscure. These include the use of bacteriostatic detergents and polymeric gel cleaners
for biocidal treatments, alternative lasers and ultrasonics for cleaning, re-purposed
traditional scagliola and mosaic techniques, grouted anchors, and several structural
reinforcing and strongback designs.
The authors will describe their techniques, review advantages and disadvantages,
provide guidelines on how and where they may be applicable, and assess their long-term
effectiveness if known. As these treatments are relatively obscure, relevant publications
are limited but available references are sited.
Keywords: bacteriostatic detergent, laser cleaning, gel cleaning, consolidation,
structural stabilization, unconventional
1. Introduction
Many conservators in practice are reluctant to part with commonly accepted methods of
addressing stone deterioration. This caution may stem from past treatment errors or
concerns experienced by the conservator or his colleagues in either the lab or the field.
This is a reasonable approach as it promotes reproducible, expected results, and reduces
the risks of failure and liability. However, such caution can also constrain the potential
for growth, innovation and advancement of the field and is ultimately self-limiting. Not
only does this prevent the introduction of entirely new and novel technologies, but it
also limits access to existing—and indeed, sometimes antique or traditional—
methodologies which have proven historical track records but have not become widely
accepted or understood. Regional preferences can also heavily influence treatment
selection, as several commonly applied European treatments—such as vacuum
consolidation, cleaning poultices and, to a somewhat lesser degree, laser cleaning—
remain under-utilized in North America. Over the past decade, the authors have sought
creative solutions to stone deterioration in sometimes unconventional places, including

orrowing and refining technologies from other industries as well as looking back at
traditional craft techniques for inspiration. The authors will present a selection of the
materials and procedures that they have worked with the hopes of introducing them to a
wider conservation audience.
2. Cleaning
2.1 Bacteriostatic Detergent
Uniquest CB-4 is a bacteriostatic detergent that penetrates microbiologically influenced
biofilms. Although initially developed for the purification of water wells, consultations
with the manufacturer suggested that this cleaner could be effective in removing
tenacious biological soiling in and on stone surfaces as well. The action of the
detergent leads to three major impacts: (1) the breaking apart and dispersion of the
biofilms; (2) the destruction of the microbial cells particularly when the concentrations
applied exceed 0.5%; and (3) dispersing attached materials formed by the infestation
and growth of microorganisms associated with the biofilms. In general these impacts on
the entrenched biofilms on surfaces require sufficient retention times to allow each of
the three events listed above to be completed in the appropriate sequence. In water wells
and pipelines the retention times vary but the ideal should be <strong>12</strong> to 24 hours for each
stage. This allows enough time for the CB-4 to effectively disperse the infesting biofilm
and associated concretious chemical accumulates; effectively reduce the active cells by
at least three orders of magnitude; and then disperse attached materials from within
underpinning surfaces (e.g. eroded pores into the marble).
The material was tested in sequence with other common detergents and quaternary
biocides. Cleaning was effected first with a surfactant (Triton X-100, 3% solution) then
a quaternary biocide (D/2 Biological Solution, 50% solution) followed by applications
of CB-4 (5% solution, paper poultice, wrapped in plastic sheeting for a 24-hour dwell
time). It is believed that the detergent removes surface soiling and exposes the biofilm.
D/2 penetrates and disrupts the biofilm and biomass. The CB-4 acts as a penetrating
biocide which attacks the structures formed on and beneath stone surfaces. These
structures then collapse and can be easily removed by light scrubbing/ rinsing. This
combined technique was used successfully to significantly reduce red-orange
microbiological staining on the white marble sculpture Slide Mantra (Isamu Noguchi;
Miami, FL), previously thought to be an intractable problem (Konkol et al, 2009). The
authors have also used this method selectively on the New York Public Library Lions
and elsewhere.
Figures 1, 2. Disfiguring biological staining before (left) and after (right) treatment with CB-4
bacteriostatic detergent.

2.2 Gel Cleaner
Both the dwell time and maintenance of
physical contact with the substrate without
drying are critical for the effectiveness of
CB-4, which was designed to be used in
water wells. A paper poultice contained in
plastic sheeting was only partially effective
at achieving this goal, particularly on vertical
surfaces where it tended to lose adhesion and
pull away. It relied solely on the penetrating
capacity of the CB-4 to reach the microbiota,
which proved generally effective on the
smooth surface of the Slide Mantra.
However, the highly porous Adair marble
National Law Enforcement Officer’s
Memorial (Washington, DC) exhibited
similar soiling conditions with a much more
challenging surface. Further, the paper
poultice method did not inherently remove
the killed biota afterwards, thus leaving
nutrient-laden beds which encourage re-
Figure 3. Prestor gel being pulled from the surface
of the National Law Enforcement Officers
Memorial, Washington, DC.
infestation and return of bio-soiling. Both the effective killing through prolonged contact
of the biocide with the biota, and the removal of the killed infestations within the pits,
were problematic and necessary to the treatment goals (Rabinowitz et al, 2011). This
was achieved by modifying the process through the use of Prestór Gel (CBI Polymers,
Honolulu, Hawaii, USA) as the transmission agent for CB-4 as a replacement for the
paper poultice. The product, based on a material developed to decontaminate nuclear
sites, is a polymeric gel that has the capacity to wick into microscopic pores. Unlike
other commercially available gel-cleaners, which are typically latex rubber-based and
polymerize into an elastic film, Prestor is water-based and water-soluble, and offers
much greater mechanical cleaning ability through its tenacious cling to and pull from the
surface. It readily clings to vertical and overhead surfaces, although its relatively low
viscosity necessitates careful site protection and application methods. As it dries, it
shrinks approximately 20% and forms a tough, pliable skin that encapsulates anything
contained within it. After approximately 24 hours, depending on environmental site
conditions, the gel can be peeled away and disposed. This period corresponds to the
necessary dwell time for the CB-4, meaning that it can be used to both ensure the CB-4
has sufficient contact time to effectively act upon the bio-colonies as well as remove
them afterward, potentially reducing re-soiling rates. One year after treatment the stone
of the Memorial remains unsoiled, confirming this effect. Please note that the tenacious
pull of the gel is not appropriate for friable substrates and should only be used on sound,
dense stones.
2.2 Ultrasonic Cleaning
Ultrasonic cleaning utilizes high frequency sound waves passed through a water
medium aided by a cleaning agent to effectively remove soiling, commonly used on

jewelry, industrial equipment and sensitive
electronics. The authors have also
experimented with expanding this technology
for use in the cleaning of masonry materials,
including limestone and terra cotta.
Ultrasonic cleaning is necessarily limited by
the size of a vessel that can contain the object
to be cleaned; it also requires a significant
investment in equipment (transducers).
Despite these potential hurdles, the authors
saw an opportunity for ultrasonic cleaning of
terra cotta and limestone elements from the
Four Seasons Fountain in Ames, Iowa. At
this site, the decorative fountain elements had
been marred by an accumulation of iron stains
and calcium carbonate deposits due to hard
water and deterioration of the internal
plumbing system. Typically, stains and
deposits of this nature might be broken down
through the use of strong oxidizing acids, or
mechanical means such as abrasive blasting.
These methods, while effective, can also
render permanent damage to the underlying
surfaces. Ultrasonic cleaning was explored as
Figures 4, 5. Panel encrusted with calcium and
magnesium deposits before treatment (above) and
after successful ultrasonic cleaning (below).
a way to effectively reducing this soiling without introducing damage to the substrate.
Through a sequence of testing using different ultrasonic frequencies, water temperatures,
and cleaning agents, parameters for treatment were established. It was determined that
frequencies in the range of 38-42 KHZ afforded effective but not overly aggressive
cleaning, aided by Branson OR, a proprietary, citric acid-based cleaning solution. Water
temperature was maintained at 150*F and a pH of 4-4.2. Each element was situated in
the custom-built ultrasonic tank and cleaned for successive 1-hour intervals, which
typically totaled 6-8 hours of treatment overall. After cleaning, the element was
transferred to a rinsing bath of deionized water. The net findings were that the
ultrasonic cleaning was highly successful at effectively cleaning terra cotta surfaces
without damaging the substrate, however, this cleaning method was not initially
successful at reducing the same deposits on the softer, more delicate limestone (Sembrat,
1998). Additional research will be needed to evaluate appropriate ultrasonic frequencies,
duration of cleaning cycles, and cleaning agents to effectively clean limestone and other
historic materials. It is possible that some modification of the system may allow for
more localized treatments, increasing its effectiveness.
2.3 Er/YAG Laser cleaning
Laser cleaning still remains underutilized in the US compared to its general acceptance
in Europe. It is not clear if this is due to the burden of the relative high cost of the
equipment, the lack of client demand for its use, concerns about the limits of the
effectiveness of the tool based on past experience, or a combination of factors. The tool
generally used in the cleaning of stone uses the Nd:YAG laser 1064 and 532 nm

wavelength. Its advantages, as well as limitations, are well known and it has been
successfully applied in the treatment of lighter toned stones for decades. The cleaning
action relies on the different absorption of the laser energy by darker materials than
lighter substrates (Martin and Cooper, 1998). Ideally suited to removing black inorganic
crusts from white stones, its effectiveness on dark substrates and organic soiling is
limited. Use of the erbium (Er:YAG) laser at 3 nm in the treatment of stone was based
on the theory by Adele de Cruz that laser energy at this wave length is absorbed by the –
OH molecule, meaning that ablation can occur on materials that can be saturated with
water or alcohol, regardless of the colour of the soiling or substrate. The tool has been
shown to be effective on removing overpaint, varnishes, wax, and bio-colonies on stone
substrates. Our tests showed it to be effective at removing wax from Rosso di Verona
marble without detrimental effects. It has also been successfully tested on removal of
lichen and overpaint on polychrome sculptures in testing. We are not aware of any
wholesale use of the technology in treatment despite these successful tests.
3. Consolidation
3.1 Nano-Particle Lime Consolidants
Throughout much of the 20th century, the primary consolidants commercially available
in the United States and Europe for the treatment of granular stone disintegration had
been organic polymers. However, research into nano-lime consolidants represents a
new wave of development with potential for further success and innovation. The
fundamental premise behind this technology is not new, namely that the consolidant
(calcium hydroxide) should be compatible with its host material (a calcareous stone,
mortar, or lime plaster). Applied lime washes (calcium hydroxide dispersed in water)
have certainly been used in the past, with varying success (Hansen et al, 2004). Calcium
hydroxide Ca(OH)2, when exposed to carbon dioxide (CO2) will be converted to
calcium carbonate (CaCO3, or calcite), much in the same way a mortar cures through
traditional carbonation. However, nano-lime particle solutions offer several theoretical
advantages over traditional limewashes. The nano-particles are borne in different
alcohols and are extraordinarily fine; for these reasons, it is hypothesized that the nanosols
can achieve a greater depth of penetration than traditional limewaters (Hirst
Conservation, 20<strong>12</strong>). Further, the fact that the solvent is alcohol, rather than water,
should also limit premature carbonation of the particles, and theoretically allow for a
greater deposition of material prior to carbonation (Doehne and Price, 2010). Not
surprisingly, depth of penetration and the time necessary for carbonation will necessarily
vary based on the host material, pore size distribution, and other environmental factors
which may not yet be fully understood (D’Armada and Hirst, 20<strong>12</strong>). Further, the fact
that the only commercially available nano-lime consolidant (CaLoSiL) is not distributed
in the United States further complicates the spread of this technology for the time being.
The authors have, to date, performed one field test using a nano-lime consolidant, on an
outdoor sculpture of Vicenza Stone, a whitish to pale-yellow Italian limestone.
CaLoSiL E-50 was diluted 1 part solution to 2 parts ethanol and gently mixed. The
solution was applied to the selected stone using a natural bristle brush, which allowed
reasonable control of the material and recapture of excess run-off, if needed. A portion
of the stone was pre-wet with ethanol to see if and how this affected absorption; because

more rapid absorption and less excess run-off
was observed without pre-wetting, this
became the treatment standard. Ultimately,
two cycles of treatment were applied to the
same area, 48 hours apart. Due to the
sensitivity of the material it was not possible
to remove a stone sample for compressive
strength or modulus of rupture calculations
beforehand. Little evidence has been
published thus far about the long-term
performance of nano-lime; the authors will
continue to intermittently monitor and make
qualitative observations about the condition
of the stone moving forward.
4. Structural Reinforcement
4.1 Grouted Anchors
Grouted anchor systems, in a variety of
proprietary iterations, have been used over
for more than 20 years to effectively stabilize
historic masonry structures by “stitching”
together unstable walls, particularly those
with internal voids or poorly bonded courses.
Due to concerns about the use of synthetic
adhesives for bedding anchors, the authors
have also successfully extended this
technology to the internal stabilization of
compromised stone sculptures. The goal is
to replace the epoxy binder with grouts
whose thermal expansion and moisture
transmission characteristics are closer to
those of stone. For installations other than
vertical cores that are accessible from the top,
Figure 6 (above). Core drilling a sculpture to
facilitate installation of a grouted anchor.
Figure 7 (below): A grouted anchor system prior
to installation.
this process requires bedding the stainless steel anchor within a polyester fabric sleeve,
then inflating it by injecting a cementitious grout under low pressure. The sleeve
expands with grout to fill the hole as the fabric is inflated. Some grout leaches through
the fabric, which allows the paste to adhere to the core walls while containing the
aggregates within the sock. With the sleeve’s flexibility, it is able to mold into the voids
and spaces within the sculptures, allowing it to be used to pin and reassemble fragments
while minimizing or excluding the use of epoxy. The sock also allows for installation in
conditions that would not otherwise be feasible, such as horizontal or upward cores,
where grout would simply run out of the holes if not contained (Cintec, 20<strong>12</strong>). The
system has been used to pin stones on the Barnard Statuary Groups at the Pennsylvania
State Capitol (Harrisburg, PA); the Tripoli Monument at the US Naval Academy
(Annapolis, MD); cast stone reliefs by artist Constantine Nivola at Yale University

(New Haven, CT); and on several travertine, coral stone, and Vicenza stone sculptures
and fountains at Vizcaya Museum & Gardens (Miami, FL).
The flow capacity of the grout and its ability to remain viscous and not separate during
installation are essential characteristics. Presstec grout used as part of the Cintec system
is the only grout the authors have found with these characteristics. Where there have
been concerns about possible differential compressive strength or when used to repair
narrow stone units, such as in the repair of the pedestal supporting a basin within a
marble fountain at Kykuit in Tarrytown, NY, narrow closed-cell foam backer rods have
been inserted into the holes parallel with the anchors to act as bond breakers and to
allow for and absorb some expansion, minimizing stress on the surrounding stone.
4.2 Strongbacks
Installation of internal anchors may not be feasible on sculptures that are very narrow or
where they are still attached to their bases and removal would not be warranted.
However, sculptures on building cornices and those installed with limited views of their
rear sides can be supported with exterior strongbacks. Usually these require attaching
anchors to the backs of the sculptures set as high up as possible to reinforce top-heavy
figures while remaining visually unobtrusive. The authors have observed several
sculptures in locations subject to high winds where these mounting sites have become
sites of damage due to concentration of forces at these points, which occurs if the
sculptures are not fully bedded and soundly pinned. The authors have developed
variations on the basic strongback design to address this issue.
A strongback for a sound sculpture with an adequate footing and mounting consists of a
solid stainless steel bar, approximately the height of the figure, mounted vertically
behind the sculpture, generally following the profile of the base and sculpture, which is
set into a reinforced concrete
footing. The bar is connected
to the figure with several
strategically placed
reinforcement points, located
at or above the mid-section
and middle-upper back.
Neoprene pads may be set
between flanges on the bar
and the stone sculpture to
provide cushioning. Ensuring
the rigidity of the post allows
for use of a rigid mounting
bolted to the sculpture. For
example, the sculpture of
Pomona located at Vizcaya
Museum & Gardens features a
Figure 8 (left). Example of a strongback mounted to the sculpture of Pomona,
an example of a sound sculpture secured to a new concrete footing.
Figure 9 (right): Example of a low-impact strongback in which nylon-coated
stainless steel cables were used to secure a sculpture on a roof parapet.
strongback that was welded
to an internal grouted
anchor which continued

through the base of the sculpture and was
embedded in a newly poured concrete footer
upon which the entire sculpture assembly was
positioned. Due to the frequency with which the
authors have had to implement this kind of
support in the Miami, FL area, they have worked
with a structural engineer to develop a design
which has been engineered to withstand the
Florida Building Code requirements for high
velocity hurricane winds. The system presumes
that the sculpture is itself well-supported.
An alternate design has been used in situations
where it is not possible to confirm that a
sculpture is soundly mounted. To avoid excess
loading at bolt points when some movement may
be encountered from high winds, a similar bar is
used but it is only attached to the sculpture with
surrounding cables. Coated aircraft grade
stainless steel cable is looped around the figure
at several locations and connected to flanges on
the vertical bar. On a roof top, this bar is also
braced with diagonal supports, all bolted to the
roof deck and waterproofed. The cables are kept snug against the figure but not overly
tightened. The cable shield consists of semi-transparent nylon, which can be painted if
needed to blend with the sculpture. This design is intended to support the sculpture
against catastrophic loss in a severe wind event while allowing for some rocking or
movement. It has proved effective on parapet figures, where some evidence of rubbing
of the cables against the stone five years after installation demonstrates that movement
has occurred. Ideally, the work should be hoisted and securely pinned as well as
provided with a strongback if needed but, in situations where this is not feasible, this
method may be used. In conditions where previous strongback installations have already
failed, leaving weakened and cracked stones that cannot safely be pinned unobtrusively,
it may be the best available method to prevent disastrous loss and limit threats to
persons and property. It can also be used for short term stabilization for life/safety
concerns pending more thorough restoration. A disadvantage of this strongback is its
higher visual intrusiveness than one solely mounted to the back, which is a greater
concern for sculptures at ground level than those atop buildings. However, a wellthought
out design, discreet placement of the reinforcing cables, and coating the bar and
cables with a sympathetic color of paint, can minimize the visual impact while still
lending adequate support to the figure.
5. Fills and Loss Remediation
Figure 10. Example of a sample stone mosaic
repair held in front of the host stone for
comparison.
5.1 Scagliola and Mosaic Patches
Exotic marbles, such as Cipollino (Greece/Mediterranean), Rosso Antico (Italy), Rosso
di Verona (Italy) or other brecciated varieties can be difficult to repair due to both the

intense colouring and vivid figuring of the marbles. When dimensional losses and voids
develop on such stones, typical patching techniques common in the conservation trade
which utilize repair mortars are not effectual due to the highly complex and variegated
colouration of the host stones. For this reason, conservators have had to borrow and
modify traditional craft methodologies into a suitable repair strategy.
Cipollino and Rosso Antico stones, which have multi-colored veins running parallel or
complexly knitted together, may be patched using scagliola techniques. Adapted from
interior plastering methods, this repair was used to great effect on pieces within the
collection of Vizcaya Museum and Gardens (Miami, FL), at the Biltmore Estate
(Asheville, NC), and at the University of Virginia (Charlottesville, VA). A palette of
restoration mortars (Jahn M<strong>12</strong>0), tinted as needed with mineral pigments, was mixed in
separate small batches to match the varying colors of the marble. For linear veined
stones, the appropriate mortar colors were shaped into flat briquettes and stacked with
others in thicknesses to match the colors and veining of the host stone. The stacked
briquette was sliced perpendicular to the sections and laid to align with the stone veins.
More complex figures can be replicated by laying in base colors, cutting back figured
areas, and then filling those voids with other colors. All fills are placed proud and
allowed to set before the repairs are dressed back to match the historic profile.
Major losses on brecciated marble columns at Vizcaya Museum & Gardens were rebuilt
either by creating a terrazzo-like patch or creating a mosaic matrix. On the mosaic
patches, a variety of stones comparable to the range exhibited in the existing column were
selected and broken or cut to create tesserae that matched the shapes in the original stone.
Once the void was properly cleared and prepared for repair, a colour-matched acrylicmodified
grout was laid into the void to serve as a bedding mortar and the stone tesserae
were set in it slightly proud of the profile to approximate the colour, veining, and texture
of the natural stone. Joints were grouted with the same mortar used in the bedding. Once
the grout was set, the entire patch area was honed down using a wet-polisher equipped
with a 50-grit diamond bit pad. Care was taken to ensure the patch repair was shaped to
match the round profile of the column. Final dressing of the patch was done using
sequentially finer pads up to 2000-grit. Care was taken to ensure adjacent surfaces of the
column were not abraded during this process. Any minor losses of tesserae or voids in
the grout were filled with additional material and screeded flush.
For stones that exhibited a more truly “broken” appearance, the process was modified to
one that resembled terrazzo. A stone similar to the embedded fragments was identified
and broken into pieces sized to match the original stone. These were cast into prepared
voids within a paste of a latex modified color-matched mortar. To ensure that the stone
fragment distribution matched that of the marble, it was necessary to cast the repair at
least ½” proud of the finished surface, as spacing between the fragments tended to
decrease towards the core of the patch. This too was then carefully honed back like the
other repairs.
6. Conclusion
The techniques and materials summarized above represent a cross-section of atypical
treatments to suit a wide variety of stone conservation challenges. Both antique and

cutting-edge, they remain outside the typical range of treatments selected by stone
conservators. However, the authors have found these somewhat unconventional
treatments to find applications in the field: sometimes modest, sometimes significant.
Undoubtedly, each technique will continue to benefit greatly from the dispersion of
information amongst our peers and with that, expanded testing, a broader shared
knowledge base, and refinement of the materials and methods.
References
“CaLoSiL® - Stone Consolidation with Nano-Particles” product literature, accessed
online 7 June 20<strong>12</strong> >
Cintec Anchoring & Reinforcement System product literature, accessed online 21 June,
20<strong>12</strong> >.
Cooper, Martin and John Larson, Laser Cleaning in Conservation (Oxford, UK:
Butterworth-Heinemann, 1998).
D’Armada, Paul and Elizabeth Hirst. “Nano-Lime for the Consolidation of Plaster and
Stone.” Journal of Architectural Conservation 18, 1 (March 20<strong>12</strong>): 63-80.
Doehne, Eric and Clifford A. Price. Stone Conservation: An Overview of Current
Research, 2 nd ed. (Los Angeles, CA: Getty Conservation Institute, 2010): 37.
“The Erbium YAG Laser: History of Application to Fine Art Conservation” accessed
online 21 June 20<strong>12</strong> >.
Hansen, Eric, et al. “A review of selected inorganic consolidants and protective
treatments for porous calcareous materials.” Reviews in Conservation 4 (2003): 14.
Konkol, Nick, et al. “Enzymatic decolorization of bacterial pigments from culturally
significant marble.” Journal of Cultural Heritage 10, 2 (July-Sept 2009): 362-366.
Mertz, Jean-Didier, et al “Sur l’Impact de la Dilation Differéntielle des Résines
Époxydes avec la Pierre.” Pre-prints Jardins De Pierres, 14 th jornées d’étude de la
SFIIC, Paris, Institut national du patrimoine, 22-24 juin 2011.
Rabinowitz, Mark, et al “Identification and Removal of Microbiological Fouling on
Adair Marble at the National Law Enforcement Officers Memorial, Washington, DC.”
Pre-prints Jardins De Pierres, 14 th jornées d’étude de la SFIIC, Paris, Institut national
du patrimoine, 22-24 juin 2011.
Sembrat, Joseph Jr. “The Use of an Ultrasonic Cleaner for the Removal of Scale
Deposits from a Terra Cotta Fountain,” in the proceedings of Cuba ’98: IVth
<strong>International</strong> Conference on the Rehabilitation of Building and Architectural Heritage.

REVISITING STONE REPAIRS: ESTABLISHING EVALUATION
PROTOCOLS AND REVIEWING PERFORMANCE
Dean Koga, Raymond Pepi
Building Conservation Associates, Inc., New York, New York
Abstract
Traditional methods for repairing stone have been supplemented over the last three
decades with new materials and techniques conforming to conservation standards
developed for historic buildings and monuments. Treatments for surface loss and
disaggregation have evolved, as have new materials for crack and loss repairs, but the
durability of modern treatments is not widely known beyond individual or institutional
experience. This paper will set forth evaluation criteria and protocols that were used to
revisit and evaluate treatments performed over the last 30 years in the New York City
metropolitan area. Adoption and modification by other investigators would be welcome.
1. Introduction
General methods for repairing exterior stone that have been used for thousands of
years still serve as the basic toolkit for conservators and crafts persons. Repairs to cracks
and losses have been the standard treatments, supplemented in recent centuries by
remedies for surface loss and disaggregation. The last thirty years have seen the use of
some different materials and techniques, but their performance in the field has been
mostly anecdotal; repair projects are rarely revisited with the express purpose of ranking
how the treatment performed over time. To the best of our knowledge there is no
durability measure for different repairs on the same material, or in different climates or
exposures. This paper is an effort to begin the process of collecting data for evaluation
of the effectiveness and durability of exterior stone repairs by revisiting treatments
performed over the last 30 years in the New York City metropolitan area.
The average longevity of surface repairs made with cement or lime based materials
has been established by observation – inspections of masonry facades often reveal failed
repairs for which installation dates are known. Over the last few decades, however,
prepackaged proprietary materials have been installed extensively in favor of traditional
site-mixed repair materials. Polymeric adhesives have found widespread use. Stainless
steel pins are used to stabilize broken or displaced stones, enabled by hand-held
percussion drills. Alkoxysilanes are now applied to disaggregating stone where lime
based or even organic materials had traditionally been used. But the change has not been
complete; traditional materials and techniques are still being employed.
To evaluate the durability of the recently developed materials and techniques as
well as the traditional repairs, a number of projects completed between 1983 and 2006
were reviewed, and the repairs were evaluated for efficacy, durability, and appearance.
A protocol and criteria for evaluation have been developed to establish the performance

of the repairs. The repaired stones are marbles (mostly dolomitic), limestones, granite,
and sandstones (primarily brown sandstone).
2. Developing evaluation criteria
The effectiveness of a repair over time should be measured against the original
treatment objective, which varies with the type of damage being repaired and its location
on a structure. Considered in this study are surface losses, cracks, and surface
disaggregation. In addition to having a physical function, many repairs also function
aesthetically by contributing to the re-establishment of the appearance intended by the
original architects or builders.
As considered in this study, loss repairs are generally intended to meet the
following objectives, recognizing that single repairs may have more than one physical
objective. Typical treatments are listed after the objectives:
• Prevent water from collecting on building surfaces: dutchman repair,
patching repair, retooling surface
• Re-establish load path continuity at large losses: unit replacement,
dutchman repair
• Reduce surface area subject to weathering forces: dutchman repair,
patching repair, retooling surface (small shallow losses)
• The repair itself must not become a hazard: dutchman repair, patching
repair
• Re-establish original architectural intent: unit replacement, dutchman
repair, patching repair
Crack repairs generally address related issues, with objectives for different types of
treatments, as follows:
• Prevent detaching stone segment from falling: anchoring, pinning
• Prevent water penetration into the crack: surface patching, grout or epoxy
injection
• Re-establish load path: grout injection
• Accommodate inherent movement: sealant or gasket installation
Surface disaggregation can be treated to the following ends, with treatments listed:
• Arrest the progress and re-establish surface cohesion: consolidation with
alkoxysilanes
• Reduce surface area subject to weathering forces: surface retooling (large
shallow areas) i
With identification of the repair objectives, factors for establishing the failure of the
repairs can be established. In the absence of failure factors, the repairs are considered
successful.
Loss repairs:
• Should not be loose or disbonded lest they become a potential falling
hazard

• Should remain cohesive to retain load path continuity, and to prevent
pieces from becoming a potential falling hazard
• Should not have separations (cracks) at interface with surrounding stone
• Should not have weathered differently from surrounding stone or shifted
color
Crack repairs:
• At repair to prevent falling, such as pinning, the piece should not have
moved
• Avenues for water entry into crack should not be present
• Where load path was re-established, movement of stone at or adjacent to
repair should not have occurred after repair
• Inherent movement was accommodated
Surface disaggregation:
• Surface should be cohesive, with no loose stone grains or crystals
• Surface irregularities that could collect water should be at a minimum
2.1 Inspection protocols
Repairs were evaluated at most of the projects from the ground level or roofs. When
repairs could be touched, they were tapped with a mallet while a hand was held on the
repair to detect any vibration that might indicate adhesion problems; the sounding
technique was used on dutchman repairs, patches, and crack repairs. Repairs were also
examined visually to detect edge separations and surface cracks. The presence or
absence of loose stone grains on surface repairs was detected by running fingers over the
surfaces.
Many repairs could not be touched and had to be evaluated by visual means only;
loose stone grains and vibrations could not be detected. Repairs were examined with
binoculars and digital photographs were taken with telephoto lenses. The contrast on the
digital images was increased to enhance any visible edge separation of patches and crack
repairs.
Evaluating whether mortar repairs had retained their original color was aided by
color photographs taken at the time of installation. The color of the mortar repair in the
original photograph was compared to the surrounding stone. The contemporary
photograph of the repair was then examined and the two photographs were compared to
detect any change in the visual difference between the mortar and stone. The
photographs manipulated for crack visibility could not be used to make the color
evaluations.
3. Repair Evaluations
Seventeen buildings or structures were visited for this study. The stone repairs in
most of the reviewed projects were designed by Building Conservation Associates
(BCA) or by the authors when they were with other firms. At all but one of the
projects ii , implementation of the repairs was overseen or monitored by BCA staff or the
authors, who also took project photographs documenting the repairs. Those

photographs, as well as the original project drawings, specifications and product
submittals were consulted for this study. When drawings were available, they were
taken to the building to locate repairs. The full information listed above was available
for only a handful of projects. In the case of the three projects from the early 1980s only
the authors’ memory was available to identify the type and location of repairs.
Completion dates for the projects ranged from 1983 to 2006.
At Brooklyn Borough Hall, one of the 17 reviewed buildings, repairs overseen by
BCA were made to the Tuckahoe marble (dolomitic) in 1989-1990. In 2001 BCA
examined the stone again and designed repairs that were completed in 2006. The 2001
survey identified some repairs from 1990 that had failed, as well as conditions that had
either not existed in 1990 or had not been addressed at the time. The building was
revisited for this study.
For this study, repairs were reviewed in the following stone types (with number of
projects in parentheses):
• Granite (5)
• Brown sandstone (4)
• Ohio sandstone (1)
• Limestone (4)
• Tuckahoe marble (4)
• Serpentenite (2)
The repairs examined were:
• Dutchman repairs, with and without metal anchors
• Cast stone veneer full face dutchman repairs
• Mortar patches and surface crack repairs: site mixed portland cement and
lime mortars, with and without acrylic admixtures; proprietary mortar
mixes, with and without acrylic admixtures; site mixed epoxy mortars;
pigmented polymeric adhesives
• Pinning repairs to cracked stones
• Injection crack repairs: epoxy; grout
• Surface retooling
• Chemical consolidation
Dutchman repairs ranging in size from roughly 100 sq. cm to 1,000 sq. cm were
examined in granite, brown sandstone, Ohio sandstone, Indiana Limestone and
Tuckahoe marble. Most were set with acrylic modified mortar; a few were set with
unmodified portland cement mortar. With one exception, the repairs employed stainless
steel anchors fabricated from stainless steel wire or rods set in epoxy or polyester
adhesive. Some of the dutchman repairs performed in 1990 at Brooklyn Borough Hall
were installed without metal anchors or ties, were found to be loose when examined in
2001, and were pinned in 2005. All dutchman repairs examined in this study were
sound. There was no evidence, either visual or tactile that any of them had become
detached or even partially separated. Edge separation of hydraulic lime perimeter mortar
joints was observed at two dutchman repairs in Tuckahoe marble, although the repairs

were otherwise sound. Some perimeter mortar joints on the 1990 dutchman repairs at
Brooklyn Borough Hall were discolored but sound.
Cast stone face veneers were employed in two rock-faced brown sandstone projects
(1983 and 1999) revisited during this study. In both cases molds for the veneers were
taken from the existing stone. The veneers were anchored with threaded stainless steel
rods set in modified polyester adhesive. None of the examined repairs exhibited any
signs of failure or apparent change in color.
BCA encounters existing patch and crack repairs of mortar, presumably without
polymer additives, in most of its masonry projects. Sound and failed repairs of both
types are usually found on the same building or structure. The failure of individual
repairs of a given type is dependent on the microclimate and individual craftsmanship of
the item as well as the inherent properties and suitability of the technique and materials.
Sound repairs are often left untreated on the same buildings where failed patches are
replaced. At Brooklyn Borough Hall, one older patch, left in 2001 as sound, is now
starting to separate at its edges.
The present study reviewed patch repairs on limestone, sandstone and Tuckahoe
marble executed with polymer modified mortars, either site mixed or manufactured, or
proprietary mortars with no polymer additives (as claimed by the manufacturer). Dozens
of proprietary patching mortar repairs were observed or inspected and failure was
observed in only one case in a repair installed in 1995. A crack had propagated through
the repair (in brown sandstone) and adjoining stones; sounding indicated that the mortar
had partially separated from the underlying stone, although no edge separation was
observed. The crack could have occurred after reconstruction of an adjacent staircase
redistributed the loads in the wall.
Because the crystalline mineral makeup of granite differs so markedly from the
surface of mortar, grains of crushed granite were set in the surface of its repair material
in all of the projects reviewed. In two projects the mortar contained acrylic additives and
the repairs remained sound, although one repair was significantly darker than the
surrounding stone. A construction era photograph of the repair was not available for
comparison. In another project, completed in 1993, epoxy was the matrix for the granite
particles. Some of the repairs appeared darker than the surrounding stone and in others
the visible portions of the epoxy matrix appeared to have yellowed. One repair on an
arris at pedestrian level was missing altogether. Remnants of the epoxy were visible in
the cavity. In the same project, detached granite fragments were reset with epoxy
mortar, which was also used to fill the shallow adjacent losses. White discoloration was
visible, as well as yellowing.
In two projects completed c. 1984, composite patches in brown sandstone were
shop or site mixed and incorporated acrylic additives in the scratch coat and the colored
finish coat. Aggregate was selected to match the color of the stone. The surface paste of
the patches has weathered somewhat, exposing the aggregate. Edge separation is
apparent on one of approximately one dozen patches reviewed.

Patching repairs performed with pigmented polymer adhesives, primarily used to
patch interior marble, were observed on serpentenite on two buildings. Polyester
adhesive was installed on one building in 1998. Of the approximately two dozen repairs
reviewed, two exhibited minor edge losses. Pigmented epoxy adhesive was installed at
the other building in 2002. Approximately twenty repairs were reviewed, and three
failures were observed, all adjacent to doorways. There was some edge loss in two
repairs. One repair had cracked through its center and had partially detached. The
original patches were pigmented to correspond to the surrounding stone, which varied
significantly and the patches blend with the stone with varying success. Construction era
photographs were not available so color shifts in the patches could not be determined.
Mortar repairs in cracks yielded varying results. The most successful repairs were
routed out a minimum of 3 mm wide and filled with site mixed acrylic modified mortar.
Edge separations were observed in repairs effected with all other types of mortars, but
not on every building. Scaffolding was available for inspection of crack repairs of
unmodified pointing mortar in Tuckahoe marble, installed c. 1979; neither the authors
nor other BCA staff were involved with the original work. The majority of the repairs
exhibited edge separation and cracks across their width. (Fig. 1) This finding supports
the widespread belief that conventional mortar repairs have a service life of
approximately 30 years. Cracks repaired with proprietary patching mortars fared much
better, although none of them had been in service as long. The failure rate was much
lower and was roughly correlated with crack width and location relative to vertical
mortar joints. Mortar in cracks narrower than 2 mm was sometimes missing. Those
types of narrow cracks were generally filled with a grout or epoxy, and the mortar was a
topping or cosmetic cap that did not contribute to the function of the crack repair.
In two instances mortar in a crack spanning between two vertical mortar joints had
separated from the stone on one edge, as had the mortar in the joints above and below.
In one case, not designed by BCA, the mortar was a hydraulic lime pointing mortar; in
the other a polymer free proprietary mortar was installed. It is probable that no mortar
would have remained intact in those circumstances. Accommodation of inherent
movement is an objective of crack repairs, but it is not always possible to know if
movement that caused a crack will recur. In the cited cases that objective was not met.
In other locations on the project where the hydraulic lime mortar was used, completed in
2003, narrow crack repairs iii were treated as sacrificial, with the expectation that an
annual or biannual conservation program would replenish the treatments. The cracks
were filled with dispersed hydrated lime putty and topped with hydraulic lime mortar,
with which the joints were pointed. When observed this year, mortar was missing from
many of the cracks.
In one limestone building, some wall sections exhibited multiple vertical crack
systems, including continuous step cracks in mortars joints. At least one of the vertical
crack systems in each wall section was treated with elastomeric sealant in the vertical
joints and lead caulking in horizontal joints to accommodate movements. The remainder
of the cracks were filled with patching mortar and the repairs are intact. (Fig. 2) The
project was completed in 2005.

Pinning repairs of cracked stone units were observed on marble, limestone and
sandstone. Approximately one quarter of the repairs could be touched and sounded, and
no movement was detected. The other repairs were examined visually with signs of
movement apparent in only one repair out of approximately three dozen. In that
instance, a marble lintel, cracked vertically through its approximate center, was
stabilized with stainless steel anchors installed across the crack into the stone above. The
crack was filled with proprietary patching mortar and has partially separated. Because
the repair objective was to stabilize the stone, and water cannot dwell in the crack, the
crack filler is performing an aesthetic function and should not be considered a failure.
Nine other repairs of the same type were observed on the building with no visible
separation at the cracked patch.
In one project completed in 1993 proprietary grout was injected into cracks in
limestone and filled them to the surface. No distress was observed in the repairs and the
color of the grout was close to that of the limestone. In another limestone project,
completed in 2002, some cracks were injected with epoxy adhesive. Where the crack
was approximately 1 mm wide or less, the epoxy was allowed to fill the crack to the
surface. Where cracks, or portions of cracks, were wider, the adhesive was raked out and
proprietary patching mortar was installed at the crack surface. Failures in the topping
treatment were described above, but the overall treatment should be judged a functional
success because the respective pieces haven’t moved, and water cannot penetrate the
crack. (Fig.3)
Shallow spalls and losses were retooled, at locations not visible to pedestrians, on a
limestone project completed in 2005. No surface erosion was visible, although the
repairs could not be touched. Some of the retooling repairs were adjacent to mortar
patching and dutchman repairs, reflecting the careful selection of appropriate repairs.
(Fig. 4) Retooling of rock-faced surfaces was observed on one brown sandstone project
completed in 1999. The surfaces were relatively intact, with some flaking at the edges of
the stone lamina. In one area, the surfaces of three stones, untreated during the repair
campaign, are beginning to delaminate. On the same building, alkoxysilane was applied
to the stones of a chimney that is out of service. No loose stone grains could be felt on
the surface of the stones; other signs of surface deterioration, such as flaking or
delamination, were not observed. Stone was consolidated with alkoxysilane on four
other projects visited for this study, but access was not available to touch the stone. This
is a repair that requires tactile inspection.
4. Conclusion
This study represents a very small sample of projects that are meant to serve as the
beginning of a data collection effort. Despite the limited number of examples, general
trends could be identified. It is no surprise that dutchman repairs were the most
successful, with no observed failure and only one documented occurrence of a previous
failure, made c. 1990. Stone remains more durable than any repair material, by far.
Metal anchors and setting materials with more adhesive capacity seem to have extended
the service life of dutchman repairs. Cast stone veneers are also effective and long
lasting repair materials.

Patching repairs have been advanced by polymer additives to mortars and by
prepackaged proprietary materials. There is a definite trend away from mixing materials
on site if they can be batched in controlled conditions and sent to the site in sacks.
Quality control is enhanced with factory mixed materials, a fact that has even affected
pointing mortars. More and more contractors and designers are asking for prepackaged
mortars. Patching mortar is part of that trend, and has performed well in the study
projects.
This study underscored the importance of selecting appropriate repairs for the
specific conditions being treated. In several projects, the full gamut of loss repair
treatments were brought to bear, depending on the size, depth of loss, and location. The
overall success of mortar patching repairs was likely enhanced by their relatively small
size, which reduces the surface area subject to failure.
The study also brought out the importance of selecting the appropriate treatment for
cracks. The forces that caused the crack should be identified, if at all possible, and the
repair selected to address the cause – or the symptom if the cause has been remedied in
some other way. If the cause cannot be identified, it is often prudent to assume that it is
still present and plan the repair accordingly. Accommodating movement is often
overlooked in crack repair design, and repair failures result. There were more failures of
crack repairs encountered in this study than any other reapir type, though the amount of
failure was small. The failures partially resulted because some causes of cracks never go
away and partially because the objective of the repair was not properly identified. In the
case of the limestone building with vertical cracks, the overall crack treatment was
successful because continued movement of the façade was recognized and multiple
crack repair treatments were employed to accommodate the movement.
It is hoped that this first attempt at studying the in-service performance of stone
repairs will spur similar efforts that will contribute to a better understanding of
contemporary stone repair materials and techniques, including effects of different
exposures and climates.

Figure 1. Unmodified mortar installed c. 1979, separated at edges and cracked across width

Figure 2. The step cracks at the left and right are filled with sealant. Horizontal segments of the
step cracks are filled with lead caulking covered by a thin coat of sealant for a consistent
appearance. The crack in the center is filled with patching mortar. Contrast enhanced for visibility
of repairs.
Figure 3. Crack injected with epoxy, note filled port holes along crack. Patching mortar is mostly
missing from crack surface.

Figure 4. Three different repairs at joint intersection: two dutchman repairs below horizontal joint;
patching repair upper left, and resurfacing above and to right of patch. Contrast enhanced for
visibility of repair
All photographs by Building Conservation Associates, Inc.
i Most often used on sandstone that has exhibited partial delamination
ii BCA is inspecting a building from scaffolding where work was last performed c. 1979.
Representatives from the contractor who performed the work discussed the repairs with
BCA.
iii The width criteria for this repair is not known