Journal of Cultural Heritage 15 (2014) 391402

Available online at
www.sciencedirect.com

Original article

Locating contact areas and estimating contact forces between the

Mona Lisa wooden panel and its frame
Giacomo Goli a, , Paolo Dionisi-Vici b , Luca Uzielli a
a
b

GESAAF, University of Florence, Via S. Bonaventura, 13, 50145 Firenze, Italy

Department of Scientic Research, The Metropolitan Museum of Art, Fifth Avenue, 1000, 10028 New York, USA

a r t i c l e

i n f o

Article history:
Received 14 May 2013
Accepted 26 August 2013
Available online 23 September 2013
Keywords:
Mona Lisa
Painted panels
Mechanical constraints
Contact pressure
Contact forces

a b s t r a c t
Since 2004 an international research group of Wood Technologists has been given by the Louvre Museum
the task of analysing the hygro-mechanical state of the Poplar (Populus alba L.) panel on which Leonardo
da Vinci painted his Mona Lisa, namely verifying the appropriateness of the thermo-hygrometric conditions in its exhibiting showcase, where the microclimate is actively controlled, and assessing the potential
consequences of any hypothetical uctuation. In order to acquire data about the mechanical behaviour
of the panel, and to feed and calibrate appropriate simulation models, the team has not only set up a continuous monitoring by means of automatic equipment, but has also performed manual measurements on
the occasion of the annual openings of the showcase where the masterpiece is conserved and exhibited.
This paper reports about techniques used for estimating the forces acting between the wooden panel
and its frame (the chssis-cadre), and their location, such data being of primary importance for evaluating
the panels internal stresses. The contact forces have been calculated on the basis of the local contact
pressures, imprinted on a pressure-sensitive foil as a range of saturation values of the colour developed
in the contact areas. The forces calculated as above have also been compared with the contact forces
between the panels back face and the crossbeams pressing it against the chssis-cadre, which have been
measured by means of a load cell. As could be expected, the results from so different techniques do not
strictly coincide; however the agreement is fairly good.
2013 Elsevier Masson SAS. All rights reserved.

1. Research aims
The research presented in this paper aims to provide realistic information about magnitude and location of the forces acting
between the wooden panel on which Leonardo da Vincis Mona
Lisa is painted, its crossbeams and its frame. Such data are of
primary importance for analysing the mechanical situation of the
panel and calibrating an appropriate simulation model of deformations and stresses produced by the environmental uctuations, in
order to evaluate and optimize any measure, which could improve
its conservation.
2. Introduction
2.1. A short description of the Mona Lisa panels structure and
geometry
Approximately ve centuries ago, Leonardo da Vinci painted his
world-known Mona Lisa on a panel made of a one-piece tangential

Corresponding author. Tel.: +393290656674; fax: +39055319179.

E-mail addresses: giacomo.goli@uni.it (G. Goli),
paolo.dionisivici@metmuseum.org (P. Dionisi-Vici), luca.uzielli@uni.it (L. Uzielli).
1296-2074/$ see front matter 2013 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.culher.2013.08.003

board of Poplar (Populus alba L.) 79 53 cm, 13 mm thick, which

arrived at our age almost unaltered except minor interventions (for
further details see [1]).
Only the front face is painted, whereas the panels original wood
surface shows up on the back face.
The panel features a complex double curvature, developed
throughout the centuries under the effect of the mechanical constraints and the environmental variations to which it has been
exposed, and also inuenced by the 11 cm-long crack, running
parallel to the grain through the panels whole thickness, starting
from the upper edge, and reaching the ladys forehead, above her
right eye.
The panel is inserted in a frame (chssis-cadre) made of Oak
wood, and is slightly forced against the 7.5 mm wide rim of the
frame by means of four Sycamore (Acer pseudoplatanus L.) wood
crossbeams, which are xed by screws to the chssis-cadre and hold
the panel atter than it would be if unconstrained. Due to the longitudinal curvature of the panel usually only two of the crossbeams
(the top and the bottom one) press against it, occasionally a third
one can be in contact with the panel.
Panel and chssis-cadre are inserted in a larger wooden gilded
frame, the only visible by the public.
An exploded drawing of the assembly (painted panel, chssiscadre, crossbeams, frame), made in 2004 by the restorers in charge

392

G. Goli et al. / Journal of Cultural Heritage 15 (2014) 391402

Fig. 1. An exploded drawing of the assembly: painted panel (1), chssis-cadre with crossbeams (2), gilded frame (3). The ssure is in the upper part of the panel.
Drawing by D. Jaunard and P. Mandron, 2004 modied.

of the wooden support, is shown on Fig. 1; due to successive interventions, the cross-sections of the present crossbeams are slightly
different from those shown in the drawing.
The panel is maintained into a climate-controlled display
case, which gets opened yearly to check the conditions of the
painting.
2.2. The main studies carried out to analyze the mechanical
situation of the panel
Since 2004 an international team of scientists has been given
by the Louvre Museum the task of analysing the hygro-mechanical
state of the Poplar panel. The questions asked by the Museums
Curators were basically to evaluate the climatic specications for
the display case, assess the risk of crack propagation, suggest possible modications to the framing system, and suggest any measure,
which could improve the conservation conditions or the annual
check-up procedure. An in-depth study of the panel, including its
wooden support and the system of cracks in the paint layers, is presented in [2] and [3]. An analysis of the risk of propagation of the
ssure laying in the upper part of the panel is reported in [4].
In order to have a better understanding of the physical and
mechanical behaviour of the panel, specic simulation models were
developed and validated against measurements and monitoring of
its actual behaviour.
The measurements include:

the forces exerted by the upper crossbeam on the upper part of

the panel, being automatically measured at 20 minutes intervals
by a monitoring equipment purposely developed and adapted
(see [5]);
the forces exerted on the contact points between the panel and
the crossbeams, manually measured every year on the occasion
of the annual opening of the showcase;
three transversal proles of the panel, measured manually every
year by means of a precision comparator on few selected points;
the shape of the panels convexity, also measured yearly by means
of optical techniques: in particular the 3D surface inside and outside its frame was reconstructed by the means of stereo imaging
and light projection systems (see [6] and [7]).
A FEM model based on heat & mass transfer + hygro-mechanical
behaviour was also implemented, and calibrated by means of the
collected data (see [8] and [9]).
Work is still on going, and further data keep being collected,
both with the same techniques and with new or improved ones.
2.3. The objectives of this paper
If the panels shape were perfectly cylindrical, it would touch
the chssis-cadre only in the central parts of its upper and lower
rims. On the contrary, the complex shape of the panel makes the
contact zones quite irregular and difcult to be identied.

G. Goli et al. / Journal of Cultural Heritage 15 (2014) 391402

393

Fig. 2. a: schematic diagram indicating the zones of the rim, where the pressure-sensitive foils was applied for the measurements; b: the right upper corner (seen from
behind) of the chssis-cadre, with the pressure-sensitive foils applied on the rim; here the chssis-cadre is dark, and the whitish pressure-sensitive foils strip is wider than
the rim, which can be seen with some difculty through the translucent lm; the purple marks on the corner have been obtained by applying pressure on pressure-sensitive
foils with a ballpoint, in order to make easy the successive identication of the strips location.

Previous direct observations indicated that contact areas are not

only distributed along the upper and lower rims of the chssis-cadre,
but also a third contact area exists on the left (seen from behind)
rim, not far from the upper side (as shown on Fig. 2).
However, despite several attempts, no accurate evaluation of
the location of contact areas, let alone of contact pressures, could
be performed before this study; difculties originated not only from
the extreme care required in any handling or manipulation of the
artwork itself, but also from the presence of the barbe (crest on
the edge of the colour layer) running all around the perimeter of
the painting itself (see [1]).
Since the actual contact areas between the panels front face and
the chssis-cadre cannot be anticipated a priori, the measurements
performed until now only allowed to roughly estimate, by means
of equilibrium calculations, just the magnitude and action line of
the resulting forces acting between panel, crossbeams and frame,
but not their actual distribution.
However, the FEM model mentioned above is quite sensitive
to the magnitude and distribution of such forces, therefore a new
series of measurements was planned in order to provide more complete and detailed ones, being of primary importance in order to be

used as input in the model, for better evaluating the panels actual
internal stresses.
This paper reports about such additional measurements, which
have been performed by means of a commercially available
pressure-sensitive multilayer foil. The contact forces have been
computed on the basis of the local contact pressures, imprinted
on the pressure-sensitive foil as a range of colour densities developed in the contact areas. The results have nally been compared,
by equilibrium calculations, with the results of the manual measurements of forces.
3. Materials and methods
3.1. The reference system
In order to properly identify the points were the forces are
applied as well as for computations, the following convenient reference system was adopted (see Fig. 5):
the origin of axes is located at the internal upper left corner (seen
from behind) of the panel, approximately coinciding with the

394

G. Goli et al. / Journal of Cultural Heritage 15 (2014) 391402

corresponding internal corner of the chssis-cadre, and lays in the

plane containing the four rims of the chssis-cadre against which
the panel is pressed;
the X axis lays in the above mentioned plane, along the internal
upper edge of the chssis-cadre, with positive direction oriented
from left to right;
the Y axis lays in the same plane, along the internal left vertical edge of the chssis-cadre, with positive direction oriented
downwards;
the Z axis is perpendicular to the same plane, with positive direction oriented towards the observer.
3.2. The Pressure-Sensitive Film
The contact zones and pressure measurements were performed using a commercial Pressure-Sensitive Film (hereafter PSF)
commercially available and produced by Fuji with the name of
Prescale . For technical specications about the product see [10].
PSF is a well-established method in medical research such as dental [11] or biomechanical research [12]. According to the product
specications, the PSF can measure the pressure applied on it with
an accuracy of 10% or less. In fact the PSF is made of two complementary lms, both made of a polyester base: one is coated
with a layer of micro-encapsulated colour-forming substance, the
other with a layer of colour-developing substance; when pressure is applied, a number of microcapsules is broken according to
the pressure level and the colour-forming substance reacts with
the colour-developing substance. Red patches then appear on the
lm and their colour density is determined by the pressure level;
conversely, colour density can be used to determine the pressure
applied. Among the seven types of PSF lm available, the type
covering a measuring range between 0.2 and 0.6 MPa was chosen, based on load cell measurements reported in [1]. In fact the
appropriateness of such choice has been later conrmed, since the
measured pressure covered the anticipated range approaching but
not reaching its limits. The thickness of an individual layer of PSF is
about 100 m; therefore their total operational thickness amounts
to about 200 m.
3.3. Performing the contact pressure measurements
The contact pressure measurements have been performed in
two successive campaigns, the rst one during the showcases
annual opening in November 2011 and the second one during the
case opening in November 2012.
In both campaigns the same general procedure was adopted:
strips of PSF lm were temporarily xed on the rim of the chssiscadre by means of a few little strips of solvent-free cotton-made
adhesive tape. The PSF strips were wider than the rim, in order to
cover all potential contact areas, even if located on the rims edge.
Once the system was prepared the panel was introduced into the
chssis-cadre and gently pressed against the rim by means of the
crossbeams, reproducing its usual mounting conditions; pressure
was applied following the instructions by the PSF manufacturer,
such as the speed of load increase (from zero to maximum in
5 seconds), and the duration of maximum load (2 minutes). Finally
the panel was extracted, and the PSF strips were carefully removed,
separated in order to avoid further accidental contacts, and packed
for transportation in order to proceed to further analysis in the
laboratory.
3.3.1. The rst campaign
For the rst campaign (2011) the PSF was applied along almost
the whole length of the rims, on the four sides of the chssis-cadre.
The area covered by the PSF lm is indicated in black on Fig. 2; the
white area was not covered because visual observations and the

measurements performed in [6] had shown that no contact could

occur there.
According to the procedure usually adopted by the restorers to
reassemble them, rstly the chssis-cadre was held almost vertical and the panel was introduced into it, then the two together
were laid horizontal on appropriate spacers resting on the table,
and nally the crossbeams were temporarily applied by tightening
them with small clamps.
Unfortunately the above-described procedure produced
unwanted contacts caused by displacements and rubbing between
panel and chssis-cadre; additional contacts were also produced
by the weight of the panel (approximately 2.465 grams), having
the same order of magnitude of the contact forces. Thus multiple
contacts, developed in different moments but inevitably imprinted
together on the same PSF, made it impossible to identify the good
data. However the possible contact areas were roughly localized,
and imprinted data showed that the sensitivity of the foil used was
appropriate.
3.3.2. The second campaign
The information acquired from the rst campaign made it possible to improve the reassembling procedure during the second
campaign (2012).
Firstly, the PSF was applied only in correspondence of the contact areas identied during the rst campaign, plus an extra-length
of 60 mm per side (see Fig. 2), so the time required was denitely
reduced and the task quite easier. Secondly, the application of the
pressure was performed as shown on Fig. 3, by holding the chssiscadre vertically, which avoided the panels weight contributing to
the measured contact pressure. Finally, the panel was introduced
into the chssis-cadre and placed in contact with the rim by sliding
it horizontally along the underlying face, thus avoiding rotation
and rubbing of the panels arris against the rim, and hence the formation of undesired marks. Once the panel was in the frame the
crossbeams were put in place and held directly by hand (see Fig. 3)
during the prescribed time (2 minutes), then the panel was gently
removed.
3.4. Deriving the pressure data and the forces from the
pressure-exposed PSF
The relationship between the colour densities imprinted on
the PSF strips and the corresponding pressure values was derived
according to the procedure outlined below.
The calibration was performed by means of the calibration
chart supplied with the PSF, allowing for the relative humidity
and temperature conditions recorded during the measurement (C
calibration curve). The colour density references from the PSF technical data sheet (Fig. 4a) were converted into grey-scale (Fig. 4b) by
desaturation, and the calibration curve (colour density vs. pressure)
provided was t with a 6th order polynomial function. The pressure
impressed strips were scanned by a commercial scanner model HP
Scanjet G4010 with a 2400 ppi resolution and saved in .png format (see Fig. 4d). The scanned strips were desaturated (Fig. 4e) and
scaled to the right dimension, the contact area selected and the
contact surface determined by ImageJ software (Fig. 4f), the average
grey value determined by The Gimp software (Fig. 4 g) and the pressure determined according to the t performed on the reference
scale (Fig. 4 h). Finally, the single zones were coloured according to
the reported colour scale (see Fig. 4i and 4l).
The total force acting on each individual area was computed by
multiplying the estimated pressure by the corresponding surface
area.
Note. The procedure described above assumes that all contact
areas and contact pressures recorded by the PSF reached their maximum values in the same moment, i.e. when the crossbars were

G. Goli et al. / Journal of Cultural Heritage 15 (2014) 391402

395

Fig. 3. The improved procedure, during the second measuring campaign (2012). a: the pressure-sensitive foils was applied on limited zones of the chssis-cadre rim; b: the
chssis-cadre was maintained vertical and the crossbeams were held tight directly by hand.

completely clamped against the chssis-cadre. In fact it could be

that the contact areas and pressures changed along time in such
a way that some maximum values took place at some intermediate phase during the process of clamping the crossbeams. However
in the described context such possibility appears unlikely, and in
any case it would have been impossible to analyse the evolution of
areas and pressures along time; therefore such possibility will not
be considered here.
3.5. Measuring manually the forces between crossbeams and
back face of the panel
The forces exerted by the crossbeams on the panel have been
measured every year since 2005 by means of a procedure and a
device presented in [5]. Some further details are given in the following. The device is composed of an uni-axial load cell equipped
with a support that can be temporarily xed to the chssis-cadre by
means of a small clamp; the load cell measures forces applied to its
front end along the direction of a threaded rod, which can be used
to adjust the height of the contact point; a steel ball and a washer
ensure a centred contact between rod and panel, and a convenient
distribution of the contact pressure on the panel (see Fig. 5b). To
perform the measurement of the force acting on the contact area
between the panel and the end of a given crossbeam, the rods end
is driven in contact with the panels back, as near as possible to the
selected area; to ensure the rmness of the contact, the rod gets
moved forward until the load cell reads a limited force. The end
of the crossbar gets then unscrewed, so that the whole load gets
transferred from the crossbeam to the load cell, without any displacement of the panel. The load cell capacity was 100 N and the
accuracy 0.03%.
Such measurement is performed at all the contact points, which
are typically Locations 1, 2, 7, 8 (see Fig. 5a), since due to the convex
shape of the panel normally no contact takes place in Locations 3, 4,
5, 6; however in 2012, possibly due to a minor modication made
on one of the panels contact points, contact took place at Location
4 as well.
Note. The measuring system described above is potentially
affected by several inaccuracies, including:

the uncertain location of the actual contact zones between the

crossbeams and the backside of the panel;
the distance between the locations where the forces were actually
measured, and the actual contact zones mentioned above;
the possible inuence of the contact force between rod and panel,
and of relaxation phenomena.
In the present context such inaccuracies could not be prevented nor analysed in greater depth; they will therefore be lumped
together, and estimated as a global inaccuracy of approximately
10%, quite larger than the one of the load cell alone.
4. Results
4.1. Contact areas and pressure values for the individual areas
The contact marks obtained from the second measurement
campaign were clearly visible and denitely produced by a perpendicular force, without any lateral sliding. Therefore, we may
assume that no double or false marks were present.
The contact marks are in most cases located near the edges of
the rims; this is clearly a consequence of the convex shape of the
panel, which can seldom rest at against the whole rim.
Also, the marks are numerous, discontinuous and quite small,
which highlights the obvious unevenness of the contacting surfaces; such unevenness can be attributed to several factors,
including effects of processing methods or tools, and of the
macrostructure of wood.
The contact areas and pressure values rounded up to steps of
0.1 MPa are shown on Fig. 6.
4.2. Forces, and their action lines
Having thus determined the individual areas and the respective
pressures acting on them (assumed to be all perpendicular to the
average plane of the chssis-cadre), the resulting force was computed for each area, for each rim of the chssis-cadre, and for the
whole of it, by simple vector summing; the results are shown in the
following. Fig. 7 shows how the contact forces are distributed along

396

G. Goli et al. / Journal of Cultural Heritage 15 (2014) 391402

Fig. 4. Outline of the procedure adopted for deriving the relationship between the colour densities recorded on the pressure-sensitive foils strips and the corresponding
pressure values. a: colour density references from the pressure-sensitive foils technical data sheet; b: desaturated colour density references from the pressure-sensitive foils
technical data sheet; d: an impressed pressure-sensitive foils strip; e: a selected part from the impressed pressure-sensitive foils strip; f: the image in (e) after desaturation;
g: contact areas manually determined and selected; h: for each area the mean grey value determined; i: pressure values identied for each individual area, according to its
mean grey value; l: each area coloured according to the estimated pressure.

the rims of the chssis-cadre; each rim was divided in 10 segments,

and the total force acting on each segment was computed, shown
in a table and graphically represented by the colour of the segment.
Fig. 8 shows the locations of the points where the resulting
forces could be considered to be applied. Such locations were computed by means of the ImageJ software, as the centre of mass of the
contact areas, each area being weighted by the grey-scale level of
the desaturated scan after image inversion.
Table 1 summarizes the total forces acting on the individual
rims, and on the chssis-cadre as a whole, resulting from measurement made with PSF method.
Table 2 summarizes the forces between the panel and the crossbeams, measured manually (according to the method described in
3.5) on the same day, about 1 hour earlier than the PSF measurements.
In fact, the moisture content and the moisture gradients of the
panel are likely to change in time, and hence also its distortion
and the contact areas and forces; therefore the results reported

Table 1
Summary of the magnitudes of forces between panel and chssis-cadre, resulting
from measurement made with PSF method. Locations A, B, and C are shown on
Fig. 8.
Location

Total force

Identication, seen
from back
Force (N)

(Top,
centre)
24.6

(High, left)

(Bottom,
centre)
20.5

50.9

5.8

PSF: pressure-sensitive foils.

here are likely to change in time, depending on the surrounding

microclimatic conditions and on their variations.
4.3. Comparison between forces measured with the two methods
In the following, the two force systems will be compared:
the forces acting between panel and chssis-cadre, measured by
means of the PSF method (see 3.3), reported in Table 1;

G. Goli et al. / Journal of Cultural Heritage 15 (2014) 391402

397

Fig. 5. a: the new crossbeams installed in 2005 (crossbeams 1 and 4 are wider than the previous ones), and the ve locations where the contact force between the panel
and the nearby crossbeam end have been manually measured (although in previous instances the contact was present in locations 1-2-7-8 only, in this case a slight contact
was detected and measured in location 4 as well). The origin of the XY coordinate system is located on the upper left corner (seen from behind) of the panel, approximately
coinciding with the corresponding internal corner of the chssis-cadre; b: the device used for the manual measurement of the contact force in selected locations.

Table 2
Summary of the magnitudes of forces between panel and crossbeams, resulting from measurements made with the manual method. Locations 1, 2, 4, 7, 8 are numbered as
in [5], and are shown on Fig. 5b.
Location

Total force

Identication, seen from back

Force (N)

(Top, left)
10.0

(Top, right)
10.4

(Mid height, right)

14.9

(Bottom, left)
15.9

(Bottom, right)
3.5

54.6

398

G. Goli et al. / Journal of Cultural Heritage 15 (2014) 391402

Fig. 6. Contact areas between the panel and the chssis-cadre, and respective pressure values, indicated by the chromatic scale. To better show their shapes and distribution,
the contact areas are also magnied by three times in the rectangles with white background. Both the contact areas and the painted layer are virtual sights, since this view
is taken from the back, through the wooden panel as if it were transparent (please note that since it is seen from the back, in this gure Mona Lisa is mirrored if compared to
the original painting).

the forces acting between crossbeams and panel, measured manually by means of the load cell method (see 3.5), reported in
Table 2.
A comparison between the results of the two measurement
methods can help in evaluating their signicance and reliability,
even though the measurements are not fully comparable due to
several factors including those listed below:
for the sake of simplicity the PSF measurement was performed by
holding in place only the crossbeams 1 and 4, and hence the contact between panel and crossbeams occurred only in Locations
1, 2, 7, 8; whereas as mentioned in 3.5 the manual force
measurements had taken place in Locations 1, 2, 4, 7, 8;
the effect of gravity can be reasonably neglected for the PSF measurements, where the panel was held vertically, as in its normal

exhibition conditions. This is, however, not the case for the measurement of forces exerted by the crossbeams where it was placed
horizontally: the self-weight of the panel (approximately 2.465
grams, i.e. of the same order of magnitude as the measured forces)
certainly induces a different curvature tendency than in the vertical position, and hence different forces exerted by the crossbars.
The analysis of such inuence requires a mathematical modelling
of the panels mechanics, and is out of the scope of this paper;
a delay of about 3 hours occurred between the two measurements, which very likely implied a hygrothermal deformation
resulting in changes of the acting forces.

However, for the purpose of this comparison, the verication

that the results are of the same order of magnitude and approximately within the range of accuracy of the measuring systems

G. Goli et al. / Journal of Cultural Heritage 15 (2014) 391402

399

Fig. 7. Distribution of the contact forces along the rims of the chssis-cadre; each rim was divided in 10 segments, and the total force acting on each segment was computed,
shown in the relevant table and graphically represented by the colour of the segment. Each segment of horizontal rims is 53.8 mm long, each segment of vertical rim is
65.9 mm long. Like in previous gures, the view is from the back.

(already discussed in the relevant sections of this paper) will be

considered satisfactory for the purpose of this paper.
Apart from what has been mentioned above, the comparison
cannot be made by just comparing the total forces shown in Table 1
and Table 2 because the forces measured with the two methods
have different application points (i.e. Locations A, B and C do not
coincide with Locations 1,2,4,7 and 8), and more complete equilibrium conditions need to be veried at any time as follows.
If the panel is physically in equilibrium, the complex of all the
forces and all moments acting on it must be balanced, i.e. their vector sums must be equal to zero. For the purpose of this analysis, we

analyse the equilibrium in the vertical position, and we neglect the

force of gravity acting on the panel and the vertical forces, which
counteract it, being applied on its lower edge. Therefore we may
consider that when the panel is vertical only the two horizontal
force systems (i.e. the forces between panel and chssis-cadre, and
the forces between panel and crossbeams) act on it, and must globally be equal and opposite to each other. In mathematical terms,
equilibrium exists when the following conditions are satised (for
the sake of simplicity the panel is here assumed to be at, with
its central plane parallel to the plane dened by the rims of the
chssis-cadre):

400

G. Goli et al. / Journal of Cultural Heritage 15 (2014) 391402

Fig. 8. Locations of the points where the resulting forces could be considered to be applied; coordinates, referred to the internal upper left corner of the chssis-cadre (assumed
to coincide with the upper left corner of the panel), were computed by means of the ImageJ software. Like in previous gures, the view is from the back, as if the panel were
transparent. The grey band surrounding the painted area is the unpainted part of the panel.

the vector sum of the forces acting perpendicularly to the panel

plane equals zero (Z = 0);
the vector sum of the moments about any horizontal axis contained in the central plane equals zero (Mx = 0);
the vector sum of the moments about any vertical axis contained
in the central plane equals zero (My = 0).
Any deviation from the above equilibrium conditions will show
an unbalance between the two force systems, or rather between
their supposed magnitudes, which were obtained by the two different measurement methods; and hence any deviation will indicate
a disagreement between the two measurement systems (with the
uncertainties discussed above).
Obviously such analysis cannot indicate by itself if and how
much one of the measurement methods is better or more

accurate than the other, all the more that an additional discrepancy
is certainly caused by the above mentioned contact at Location 4
and force of gravity action; however a reasonable agreement will
add value to both of them, and encourage the exploitation of the
one providing, case by case, the information most useful and appropriate for specic tasks.
As regards condition (a), i.e. equilibrium of total forces acting along the Z axis, the forces measured with the PSF globally
amounted to 50.9 N, while the forces measured with the manual system globally amounted to 54.6 N. The resulting unbalance
amounts to 3.7 N, or 7.3% of the PSF measurement. If we consider
that:
according to the PSF specications the measurement accuracy is
10% of the measure itself;

G. Goli et al. / Journal of Cultural Heritage 15 (2014) 391402

401

Table 3
Calculation of moments acting on the panel after having chosen reference axes (x = 6 mm and y = 229 mm) which provided largest magnitude of the unbalance, and assuming
the forces measured by the load cell to be located at mid-width of the crossbeam and at 10 mm distance from the panels edges. This computation intends to quantify the
amount of the unbalance of the moments applied to the panel by the forces measured with the two measurement methods: (1) the second-last row shows the nal unbalance
values, i.e. the algebraic sum of all moments applied to the panel (both the ones produced by the crossbeam forces, measured by the load cell method, and the ones produced
by the contact with the chssis-cadre, measured by the PSF method); (2) the last row shows the same nal unbalance values, expressed as percentages of the sum of the
absolute values of the moments produced by the forces measured by the PSF method.
Moments referred to X line (parallel to X axis)
Moment arm [m]

Moments referred to Y line (parallel to Y axis)

Force [N]

Forces exerted by crossbeams, and resulting moments

0.037
10.0
Location 1
0.037
10.4
Location 2
Location 4
0.270
14.9
0.753
15.9
Location 7
0.753
3.5
Location 8

Moment [N m]

Moment arm [m]

Force [N]

Moment [N m]

0.37
0.38
4.02
11.94
2.61

Forces exerted by crossbeams, and resulting moments

Location 1
0.219
10.0
Location 2
0.299
10.4
Location 4
0.299
14.9
Location 7
0.219
15.9
Location 8
0.299
3.5

2.19
3.11
4.44
3.48
1.04

Forces calculated by PSF method, and resulting moments

0.000
24.6
Location A
0.225
5.8
Location B
0.778
20.5
Location C

0.00
1.30
15.96

Forces calculated by PSF method, and resulting moments

Location A
0.146
24.6
Location B
0.223
5.8
Location C
0.000
20.5

3.58
1.30
0.00

Sum of all the moments applied [N m]

2.06

Sum of all the moments applied [N m]

0.64

% of sum of PSF moments [%]

12.0

% of sum of PFS moments [%]

13.0

PSF: pressure-sensitive foils.

in 3.5 the accuracy of the manual system was estimated around

10% as well, we may conclude that the two systems provide forces
along the Z axis having very similar magnitude conrming that
the total forces are comparable.
For conditions (b) and (c) the reference lines in respect to
which the moments should be calculated have been dened by
means of the following procedure. Since the computed moments
are not independent from the assumed reference lines, it would
be meaningless to express the resulting unbalance of moments as
a percentage of the magnitude of a generic resulting moment. In
order to express the unbalance as objectively as possible, reference lines X and Y, parallel to X and Y axes, were chosen so that
the unbalance was the largest possible. This was done by means
of a trial-and-error procedure applied to the absolute values of
the moments computed for the PSF method. The lines resulting
in the largest unbalance resulted as y = 6 mm for the line X and
x = 229 mm for the line Y, in the reference system dened in 3.1.
For the computation of the acting moments the forces measured
with the manual system were considered as acting on the centre of
the crossbeams width, at a distance of 10 mm from the panel edge.
Table 3 shows the simple calculations expressing the equilibrium
of the moments.
As can be observed in Table 3, the sum of the moments acting
on the X line amounts to 2.06 N m (or 12.0% of the moment calculated by PSF method), and similarly the sum of the moments about
the Y line amounts to 0.64 N m (or 13.0% of the moment calculated by PSF method); hence the two methods provide comparable
moments, being very close to the systems estimated accuracy.
5. Conclusions
The PSF method for measuring the contact areas and contact
pressures between panel and chssis-cadre, and hence computing
magnitude and action lines of the contact forces, has here been
implemented and evaluated. The following conclusions may be
drawn from the reported tests and calculations:
the production of the contact marks is totally non-invasive, and
can be performed in a reasonably simple way. However, it is
essential to implement appropriate well planned procedures,
namely in order to prevent the formation of false marks, produced by slipping between the contacting surfaces;

the processing of the imprinted marks in order to derive the

pressures, the forces and their action lines is complex and
requires several steps and calculations; the whole procedure can
be performed by means of open-source software, however some
steps could be made quicker and simpler by using specic commercial software;
the reliability of the PSF method has been tested by comparing
its results with those from the manual measurement of forces
acting between panel and crossbeams by means of a load cell.
Obviously such comparison cannot indicate by itself if and how
much each method provides true results, or one of the methods
is better or more accurate than the other; however the resulting
reasonable agreement supports the validation of both of them,
and encourages the exploitation of the one providing, case by
case, the information most useful and appropriate for specic
tasks;
the results obtained by the PSF method, which provides both
magnitude and location of the contact forces, can be used as
a valuable input in mathematical hygro-mechanical models for
the prediction of the mechanical response of the panel, under
changing micro-environmental conditions.

Acknowledgments
The authors would like to thank the numerous colleagues who
made this paper and the described work possible. Among them,
unfortunately too many to be all cited, the following ones are specifically mentioned with pleasure and gratefulness:
Vincent Delieuvin, curator in the department of paintings, Muse
du Louvre, for giving permission to perform this new kind of test;
Elisabeth Ravaud, from C2RMF, coordinator of the testing schedule, for allocating the necessary time slots;
Daniel Jaunard and Patrick Mandron, restorers of wooden supports, for their contribution in dening the procedure for
obtaining reliable contact marks and also for providing their
drawing shown on Fig. 1;
Joseph Gril and all the colleagues of the study team, for the support given in deciding and implementing the described tests;
the colleagues Linda Cocchi and Paola Mazzanti for their help and
support in preparing the PSF strips and during the manual force
measurements.

402

G. Goli et al. / Journal of Cultural Heritage 15 (2014) 391402

Authors contributions: G. Goli mainly set up and carried out

the tests and analyses with the PSF lm; P. Dionisi-Vici mainly
set up and carried out the manual measurement of the forces;
L. Uzielli coordinated the measurement procedures. The three
authors equally contributed to the writing of the paper.
References
[1] J.P. Mohen, M. Menu, B. Mottin (Eds.), Mona Lisa - Inside the painting, Harry N.,
Abrams, Inc., New York, 2006.
[2] E. Ravaud, The Mona Lisas wooden support, in: J.P. Mohen, M. Menu,
B. Mottin (Eds.), Mona Lisa - Inside the Painting, Abrams, New York, 2006, pp.
3237.
[3] E. Ravaud, The complex system of ne cracks, in: J.P. Mohen, M. Menu, B.
Mottin (Eds.), Mona Lisa - Inside the Painting, Abrams, New York, 2006,
pp. 3842.
[4] J. Gril, E. Ravaud, L. Uzielli, J.C. Dupr, P. Perr, D. Dureisseix, O. Arnould,
P. Dionisi-Vici, D. Jaunard, P. Mandron, Mona Lisa saved by Grifth theory:
assessing the crack propagation risk in the wooden support of a panel painting,
in: M. Fioravanti, N. Macchioni (Eds.), International conference on integrated
approach to wood structure, behaviour and application, joint meeting of ESWM
and COST Action E35, Florence, Italy, 1517.5.06, 2006, pp. 109114.

[5] L. Uzielli, P. Dionisi-Vici, J. Gril, Physical and mechanical characterization of

the support, in: Mona Lisa - Inside the Painting, Abrams, New York, 2006, pp.
4849.
[6] F. Brmand, P. Doumalin, J.C. Dupr, F. Hesser, V. Valle, Measuring the relief
of the panel support without contact, in: J.P. Mohen, M. Menu, B. Mottin (Eds.), Mona Lisa - Inside the Painting, Abrams, New York, 2006, pp.
4347.
[7] F. Brmand, P. Doumalin, J. Dupr, F. Hesser, V. Valle, Relief analysis of the
Mona Lisas wooden panel, in: Proceedings of the XIth International Congress
and Exposition of the Society for Experimental Mechanics, Orlando (FLORIDA),
2008.
[8] D. Dureisseix, J. Gril, O. Arnould, Mechanical modeling of the activity of the
exible frame, in: J.P. Mohen, M. Menu, B. Mottin (Eds.), Mona Lisa - Inside the
Painting, Abrams, New York, 2006, pp. 5051.
[9] P. Perr, R. Rmond, J. Gril, Simulation of the effects of ambient variations, in:
J.P. Mohen, M. Menu, B. Mottin (Eds.), Mona Lisa - Inside the Painting, Abrams,
New York, 2006, pp. 5051.
[10] Fujilm,
Prescale
lm,
http://www.fujilm.com/products/prescale/
prescalelm/ [site accessed on May 2013].
[11] D. Koc, A. Dogan, B. Bek, Bite force and inuential factors on bite force measurements: a literature review, Eur. J. Dent. 4 (2010) 223232.
[12] R. Allaire, M. Muriuki, L. Gilbertson, C.D. Harner, Biomechanical consequences
of a tear of the posterior root of the medial meniscus similar to total meniscectomy, J. Bone Joint Surg. 90 (2008) 19221931.