Current Approaches to

the Structural Conservation

of Panel Paintings
P A R T F O U R

306
M
ort naxaot was caustn by the great ood of 1966 in
Florence than by both World Wars combined. Many paintings
and other artifacts were submerged in the oodwaters for more
than eighteen hours. They were covered with mud mixed with heavy
deposits of heating oil that had seeped from the storage tanks housed in the
many basements of the city. The worst damage was done to the large num-
ber of panel paintings in Florence and the surrounding countryside; those
that had been submerged swelled many inches beyond their original size.
Subsequently, these paintings were subjected to a long and gradual
drying process, rst in the limonaia, the old hothouses built by the Medici
in the Boboli Gardens for their favorite collection of citrus plants. These
hothouses were quickly converted into one large humidity chamber. The
humidity was raised to 95% at a temperature of 12 C over a two-year
period. Afterward, the treatment was continued in the former army bar-
racks of the Fortezza da Basso, which in the meantime had been trans-
formed into the largest restoration laboratory in the world; it had, in fact,
become an independent governmental department, a soprintendenza, by
special decree.
Despite the carefully controlled drying process, many of the panels
shrank considerably. This shrinkage caused severe blistering and cupping of
the paint layers, as well as deformation of the supports (Cianfanelli, Ciani
Passeri, and Rossi Scarzanella 1992). Consequently, many of the panel
paintings had to be transferred to canvases and to new, rigid supports. The
oil deposits were removed with a poultice made from Shellsol A and talc
applied to a Japanese-tissue interleaf.
The devastation caused by the ood was, to some degree, oset
by the benet of the better understanding that was gained about the
behavior of wooden artifactspanel paintings in particular. For instance,
the negative eects of dovetails, which had already gone out of style by
the end of the 1950s, were conrmed (see Rothe, Critical History,
herein). The negative eects of rigid restraints or crossbars in relation to
the natural exibility of panels were better understood. It became clear
that those restraints that held the panels in place but did not hinder their
need to expand and contract were the most eective.
It also became obvious that the materials that were used for
crossbars had to be stable and unaected by environmental uctuations.
Mansonia, which had been widely used in Florence by the restoration
Florentine Structural Stabilization Techniques
Andrea Rothe and Giovanni Marussich
departments of the Vecchie Poste at the Uzi and Palazzo Pitti before
the ood, proved to be the most stable wood, with the least tendency to
deform (see Rothe, Critical History, herein).
1
Used for more than forty
years in the construction of crossbars, mansonia functions very eciently
and, in fact, appears to be better than any other type of wood because of
its density and workability. Panels with mansonia crossbars expanded and
contracted drastically after the ood but did so with little or no buckling.
Planks of mansonia that had been immersed for over a week and then
inadvertently used as gangways to wheel mud out from the ground oor
of the Vecchie Poste did not deform or crack, and they were later utilized
to make new crossbars. Today mansonia is still usedalthough much less
often because of its toxic properties. Other woods, such as steamed beech,
have also been used but have not given such satisfying results. Metal cross-
bars, such as those used successfully in Rome by the Istituto Centrale del
Restauro, have rarely been used in Florence, primarily because of aesthetic
considerations (see Rothe, Critical History, herein).
If a panel is in good condition, the conservator usually chooses not
to intervene. Unfortunately, this is not always possible. Intervention is nec-
essary whenever the original crossbars have been lost (causing warpage),
the panel has previously been thinned, splits have caused loss of color, or
panels have cracked apart. The restraint that a brace or crossbar should
exert on a panel is dicult to measure or predict, but today the rule is to
give the panel ample lateral freedom to move and to manipulate the origi-
nal surface as little as possible by making the braces much smaller than was
formerly considered appropriate, and thus more exible (Figs. 1, 2).
Excessive restraint such as that caused by older cradles tends to
block the movement and facilitate the formation of new cracks and even
of splits (Figs. 3, 4). Conversely, too little restraint can allow panels to
deform, especially those that have been thinned and have lost their original
coating (see Rothe, Critical History, herein) or the aged skin that
307 Fiortri t Srrucrurai Srati ii zari o Ttcni guts
Fi gure 1
Guglielmo di Pietro de Marcillat,
Annunciation, 1524. Reverse. Oil or mixed
technique (?) on panel, 180 150 cm.
Convent of S. Francesco, Sargiano, Arezzo. A
typically heavy crossbar of the early 1970s,
with pegs glued and screwed to the panel; a
wide swath of original wood surface was
removed to create a level area. The crossbar
on the bottom is original.
308 Rot he and Mar us s i ch
Fi gure 2
Examples of crossbars showing progressive
reduction in size. (a) Crossbar used in 1975 on
the Annunciation by Guglielmo di Pietro de
Marcillat, Convent of S. Francesco, Sargiano,
Arezzo; the panel is 150 cm wide, the crossbar
7.5 cm wide. (b) Crossbars used in 1988 on
the Nativity by Girolamo di Benvenuto, The
J. Paul Getty Museum, Los Angeles; the panel
is 161 cm wide, the crossbars 4.5 cm wide.
(c) Crossbars used in 1989 on The Birth of
Bacchus by Giulio Romano, The J. Paul Getty
Museum; the panel is 80 cm wide, the cross-
bars 3.2 cm wide. (d) Crossbars used in 1987
on The Card Players by Joos van Crasbeeck,
The J. Paul Getty Museum; the panel is
31.1 cm wide, the crossbars 2.7 cm wide. (e) If
crossbars were to be placed on The Card
Players today, a smaller version (1.8 cm wide)
would be used. (f ) Crossbars used in 1990 on
The Abduction of Proserpine by Alessandro
Allori, The J. Paul Getty Museum; the panel is
228 cm wide, the crossbars 3.3 cm wide.
e
d
c
b
a
forms on the back of old panels (consisting primarily of compacted wood
cells and accumulated dirt).
In the Florentine approach to rejoining panels, the precision with
which the work is carried out is key to the success of the treatment. This
approach is described as risanamento delle tavole, making panels sound
again. The pivotal task is to cut precise V-shaped grooves of approxi-
Fi gure 3, bel ow
Girolamo di Benvenuto, Nativity, ca. 1500.
Reverse. Tempera on panel, 204 161 cm.
The J. Paul Getty Museum, Los Angeles. A
heavy cradle is glued to the panel, which had
been thinned to less than 12 mm. This inter-
vention dates to about 1900.
Fi gure 4, above
Girolamo di Benvenuto, Nativity. Detail. This
raking-light photograph shows distortions and
cracks on the surface caused by the thinning
of the panel and the restraint of the heavy
cradle.
f
mately 55. The groove should straddle the crack all the way down to the
gesso preparation; short, individually tted wedges are then inserted into
these grooves. The grooves should be made as deep as possible without
causing damage to the paint layer, so as to avoid the formation of hairline
ssures (see Rothe, Critical History, herein).
The type of wood used to reconstruct these panels should be well-
aged material of the same type as the original painting support. The vari-
ous chisels used, including a pointed chisel for the nishing of the V-shaped
grooves, must be maintained in constant sharpness (Fig. 5). If percussion
is needed, the ball of the hand (never a mallet) may be used. In some
instances, when the cracks are straight and long, two angled planes are
usedone for the left side of the split, the other for the right side (Fig. 6).
Before the wedges are inserted, the detached sections of the panel
must be perfectly ush with each other. This is accomplished by a simple
system of temporary braces, or tiranti, that are screwed into the panel
309 Fiortri t Srrucrurai Srati ii zari o Ttcni guts
Fi gure 5
Some tools used in the preparation of
V-shaped grooves.
Fi gure 6
Two angled planes, sometimes used for
preparing long, straight grooves.
wherever necessary along the crack. By strategic placement of the screws
and the small blocks under the braces, either side of the split can be
pushed down or pulled up (Fig. 7). If the panel is very thin, little blocks of
wood can be temporarily glued onto the panel to hold the screws in the
areas that need to be leveled. The glue used for softer woods, such as
poplar and limewood, is mostly a polyvinyl acetate (PVA) emulsion glue
such as Vinavil, thinned with water.
2
Woodworkers point out that the
glue that oozes out is what ensures a lasting bondmeaning that the less
glue that remains between the wedge and the wood of the panel, the bet-
ter. For harder woods such as oak, a two-component epoxy glue such as
Araldite is used.
3
For those who are not master artisans, a simpler and quite
eective method was developed by Barbara Heller at the Detroit Institute
of Arts after she worked for many years in Florence (Heller 1983). She cuts
the grooves with a router and uses precut V wedges that are set in with
Araldite carvable paste.
4
The results have been very encouraging and seem
to be stable, especially in the case of softer woods such as poplar.
The movable crossbars are held in place by pegs, or nottole, that
are glued to the panel with an epoxy adhesive. The section of the cross-
bars is trapezoidal, and particular care is used in planing the sole and the
two side edges. To ensure a perfect glide, hot paran is applied to the
edges and polished, and the same is done to the face of the pegs.
The crossbars and pegs of the early 1950s were much heavier and
wider. The pegs were not only glued to the panel but also screwed on,
thus locally blocking the movement of the panel. Two or three wide
swaths were also planed at across the panel to accommodate the width
of the crossbars with the pegs (Fig. 1). This method removed much of the
aged skin, something that is no longer done today. To overcome the irreg-
ularities of the panel, individual spacers are now tted and glued between
the pegs and the panel.
The Opicio delle Pietre Dure restoration department at the
Fortezza da Basso has carried out more panel restoration than any other
institution in the world; consequently, it has gained a wealth of unique
experience. It has introduced and perfected many new systems that reduce
interference with the tendency of wood to move. Where deemed appro-
priate, the angle of the V-shaped cuts has been reduced at times from 55
to just 7.5 with a special router bit (Castelli, Parri, and Santacesaria 1992).
Although this approach interferes less with the original wood, the wood-
worker does not have as much control with a router as with a handheld
chisel and therefore cannot cut as close to the original gesso layer; this
deciency might, in time, result in a weaker joint (Castelli, Parri, and
Santacesaria 1992).
Other systems may be used to minimize the interference with
the original panel, such as the method of attaching the crossbars without
pegs. Instead, a system of sparsely distributed brass threaded inserts is
screwed and glued into the panel. The crossbars are slotted lengthwise at
the same intervals as those of the threaded inserts, and identically slotted
brass plates are set into the crossbars. These crossbars are then attached
with long bolts that t into the center of the slotted brass plates and are
directly screwed into the threaded inserts glued into the panel. The bolts
are not tightened excessively, and a Teon washer can be used to facilitate
lateral movement. A simplied version of this method consists of fasten-
ing the crossbars, which are also slotted, with long, round-headed brass
310 Rot he and Mar us s i ch
Fi gure 7
Temporary tiranti, or levers, used to pull dis-
torted surfaces of separated panel boards back
into plane before wedges are glued into previ-
ously prepared grooves.
screws that are inserted directly into the original wood of the panel.
Unfortunately, if the crossbars need to be removed and reattached several
times, the screw holes will eventually wear out if this simpler method is
used. In either case, to prevent rusting, only brass screws and steel bolts
are used (Fig. 8).
At times panels need to respond in more than one direction to
humidity uctuations. Expansion and contraction are sometimes aug-
mented by a tendency of the panel to warpa tendency that, if impeded,
might cause the panel to split. For this reason methods have been devised
to add some form of spring action to the construction of crossbars. The
simplest method consists of adapting existing older cradles with springs
that are tted into carved recesses at the junction of the braces and bat-
tens. For this purpose the battens are also thinned to facilitate movement
(Castelli, Parri, and Santacesaria 1992).
Another method improves the system of bolts discussed above in
the construction of new battens or the adaptation of original ones. It con-
sists of steel springs of approximately 2.5 7.5 cm that are lodged into
slotted and carved recesses in the crossbars so as to give the bolts ample
space to move and to allow the panel not only to expand and contract but
also to ex up and down (Castelli, Parri, and Santacesaria 1992). A more
sophisticated method makes use of conical springs that are inserted into
the crossbar. The brass nuts are held in place by pegs made out of lime-
wood glued to the back of the panel (Castelli, Parri, and Santacesaria 1992).
A system for thin panels that provides the most freedom of move-
ment consists of a strainer that is constructed around the panel. The
strainer holds the panel in place with springs attached to small wooden
blocks that are glued to the panel. This system is not ideal for environ-
ments that have no climate control, as it does not oer enough restraint to
the panel: in some cases panels treated in this manner have deformed and
cracked. A much simpler and more eective solution in this case is the
mounting of the painting into its frame with steel springs, as has been
done at the Bavarian State Galleries in Munich (by Christian Wolters).
5
The newest methods, which are mentioned by Castelli (see Restoration
of Panel Painting Supports, herein) deal with more sophisticated spring
mechanisms that permit panels to ex.
The guiding idea behind all these constructions should be to give
the panels ample room to move while at the same time exerting a certain
amount of restraint to keep them from deforming. The authors have
observed old panelssuch as a painting by Lorenzo Sabbatini, Madonna and
Child Enthroned with Two Saints from the Staatliche Museen zu Berlin (Bode
Museum)that have deformed because they have lost all or part of their
original restraints (Fig. 9). The general guideline is not to treat a panel if it
has survived in good condition, but if original crossbars are missing and the
panel has a tendency to deform, the crossbars need to be replaced. Wooden
panels need to be held in plane gently but rmly; otherwise they may
deform, especially if exposed to uncontrolled climatic environments, as is
the case with the great majority of panel paintings in the world.
Moisture barriers can be of some help in the centuries-old drying
process of a panel by slowing down its constant response to changes in
humidity (Buck 1978). The most commonly used materials have been
Lucite 2044 or 2045 and Acryloid B72.
6
Saran and wax have also been used.
7
Fortunately, the unaesthetic and sometimes heavy constructions of wax
and balsa wood that have been used often in the United States and England
311 Fiortri t Srrucrurai Srati ii zari o Ttcni guts
Fi gure 8
Various screws and bolts used in less invasive
types of crossbar attachments.
Fi gure 9
Lorenzo Sabbatini, Madonna and Child
Enthroned with Two Saints, ca. 1560. Reverse.
Oil on panel, 151.8 229.0 cm. Staatliche
Museen zu Berlin, Preussischer Kulturbesitz,
Gemldegalerie. The loss of the central cross-
bar has caused severe convex warpage. The
dovetail insets are probably original.
have rarely been adopted in Florence. New problems for future interven-
tions are created when materials such as wax cannot be removed com-
pletely; their residues can prevent the eective use of PVA or epoxy glues.
Some previous attempts to straighten poplar panels were made by
thinning them down to less than 7 mm and attaching heavy cradles to the
backs, as in the case of the Madonna and Child with Musical Angels by
Gherardo Starnina in the J. Paul Getty Museum in Los Angeles, California
(Fig. 10). The eectssuch as severe cupping or aking of the paint lm
of these radical interventions can often be seen on the front of the paint-
ing (Fig. 11). In cases in which the original support has been severely
altered, it may actually be benecial to attach the panel instead to a rigid
support, such as a laminated strip board, rather than to let it move freely,
as previously described (Fig. 12). For example, a painting attributed to
Giovanni Bellini, The Presentation in the Temple (private collection, Venice),
which has a severe aking problem, had been thinned to less than 5 mm
and cradled. It was decided to attach the panel painting to a laminated
strip board after it was evenly planed to a thickness of about 4 mm. The
glue used was Vinavil, a PVA emulsion, although today (as was used on
the Starnina) an epoxy adhesive such as Araldite is preferred in order to
avoid the excessive absorption of water from the PVA emulsion. Since
treatment in 1966 the Bellini has been exposed at various times to a com-
pletely uncontrolled environment (Tintori and Rothe 1978). As with the
312 Rot he and Mar us s i ch
Fi gure 10
Gherardo Starnina, Madonna and Child with
Musical Angels, ca. 1410. Tempera and gold on
panel, 92 x 51.3 cm. The J. Paul Getty
Museum, Los Angeles. The pronounced
cracking of the paint lm was caused by
excessive drying of the back following an
intervention of more than sixty years ago. At
that time, the poplar panel was reduced from
its original thickness of more than 25 mm to
less than 5 mm.
313 Fiortri t Srrucrurai Srati ii zari o Ttcni guts
Fi gure 11
Gherardo Starnina, Madonna and Child with
Musical Angels. Detail. This raking-light photo-
graph of the upper left portion shows pro-
nounced cracking of the paint lm.
Fi gure 12
Gherardo Starnina, Madonna and Child with
Musical Angels, reverse. After the back of the
painting was planed even, it was attached to a
laminated strip board with an epoxy adhesive
(see note 4). This method also creates a
humidity barrier.
Starnina panel, the treatment of which was carried out in 1982, the condi-
tion is stable, and no new signs of cupping or aking have been observed.
8
The conservator must always keep in mind where objects are to
be housed. In a climatically stable environment, even a heavy cradle will
have very little negative eect on a painting; consequently, it might be wiser
to leave well enough alone. Many paintings, however, must be returned to
environments that are not climate controlled. These paintings need ade-
quate freedom of movement, some form of moisture barrier (without
complex constructions), and protection from structural experiments. New
methods and ideas are constantly being developed, and though it is in the
nature of conservators to continually change, one sometimes cannot help
but wonder if is not better to stay with some of the structural conservation
methods that have proved their eectiveness over time, rather than con-
stantly expose panel paintings to experimental innovations.
1 Mansonia altissima; the tree comes from the rain forests of Ghana, Ivory Coast, and Nigeria.
The sapwood has characteritics similar to those of the heartwood; the heartwood, which is
slightly toxic, is most often used.
2 Vinavil NPC, Stella Bianca, is a nonionic dispersion of medium plasticized acetate emulsion in
water (see Materials and Suppliers).
3 General-purpose epoxy structural adhesive AW106 and hardener HV 953 (see Materials
and Suppliers).
4 Epoxy structural adhesive (carvable paste, wood) AV 1253 and HV 1253 (see Materials
and Suppliers).
5 Wolters has also supplied informationverbally and by demonstrationabout this type of
mounting (Munich, 1956).
6 Lucite 2044 and 2045 are the Italian product names; in the United States they are also called
Elvacite. The adhesive 2044 is an n-butyl methacrylate, and 2045 is an isobutyl methacrylate.
Both are of high molecular weight. Acryloid B72, also known as Paraloid B72 in Europe, is an
ethyl methacrylate copolymer. (See Materials and Suppliers.)
7 Saran F.120 is a vinylidene chloride-acrylonitrile copolymer. It was rst introduced by
Richard Buck in 1961. After the ood, Sheldon Keck came to Florence and proposed a 30%
solution in methyl ethyl ketone as a moisture barrier. Saran F.220 was also used. (See Materials
and Suppliers.)
8 Both treatments were executed by Giovanni Marussich and Renzo Turchi.
Acryloid B72, Rohm and Haas Co., Independence Mall Street, Philadelphia, PA 19105.
Araldite AV 1253/HV 1253 and AW 106/HV 953, Ciba-Geigy Corporation, 4917 Dawn Avenue,
East Lansing, MI 48823.
Elvacite, Du Pont Company, Polymer Products Dept., Methacrylate Products Group,
Wilmington, DE 19898.
Saran F.120 and F.220, Dow Corning Corporation, Midland, MI 48640.
Shellsol A, Shell Oil Company, P.O. Box 4320, Houston, TX 77210.
Vinavil NPC, Stella Bianca, Enichem Synthesis, Italy.
Materials and Suppliers
Notes
314 Rot he and Mar us s i ch
Buck, R. D.
1978 The use of moisture barriers on panel paintings. In The Behavior of Wood and the
Treatment of Panel Paintings, 2636. Minneapolis: Upper Midwest Conservation
Association, Minneapolis Institute of Arts.
Castelli, C., M. Parri, and A. Santacesaria
1992 Supporti lignei: Problemi di conservazione. In Problemi di restauro, riessioni e ricerche,
ed. Umberto Baldini, 4163. Florence: Edizione Firenze.
Cianfanelli, T., F. Ciani Passeri, and C. Rossi Scarzanella
1992 Il consolidamento dei dipinti su tavola. In Problemi di restauro, riessioni e ricerche,
ed. Umberto Baldini, 89108. Florence: Edizione Firenze.
Heller, B.
1983 The joining of wood panels with Araldite epoxy system. Abstract of a paper presented
at the Paintings Specialty Group, 11th Annual Meeting of the American Institute for
Conservation, 29 May, Baltimore.
Tintori, L., and A. Rothe
1978 Observations on the straightening and cradling of warped panel paintings. In
Conservation of Wood in Painting and the Decorative Arts, ed. N. S. Brommelle, Anne
Moncrie, and Perry Smith, 17980. London: International Institute for Conservation
of Historic and Artistic Works.
315 Fiortri t Srrucrurai Srati ii zari o Ttcni guts
T
ni s arri cit rrtstrs work by the Division of Restoration for
Canvas and Panel Paintings at the Opicio delle Pietre Dure e
Laboratori di Restauro (OPD) in Florence.
The paintings described below were selected because of their varied con-
struction techniques and the conservation problems they pose, problems
that were not remedied by past restoration attempts. Presentation of these
works provides an opportunity to explain various options for the repair,
consolidation, and construction of support and control systems for panel
paintings. Eective examples of restoration have in common critical
methodologies that oer the least possible invasion of the artwork. All the
original components of the work are respected. It is understood that every
intervention to the wooden support entailing alterations, intrusions, or
substitution of support parts or of the control structures may give rise to
dangerous, dicult-to-control tensions and deformations in the wooden
construction.
Interventions were tailored for each painting with the aim of
designing a coordinated restoration plan that addressed each panels particu-
lar problems. To prepare for such a plan adequately, the data-gathering
phase in conservation is fundamental.
Understanding a work of art begins with the study of its original
construction technique, the state of preservation of all its constituent
materials, and any past restorations. Subsequently, the conservator should
select appropriate diagnostic tests that deepen this understanding and
assist in identifying past conservation attempts. Finally, the conservator
can outline a plan for the various restoration phases.
For the design of the plan, it is imperative to know the relative
humidity (RH) of the environment from which the painting came, as well
as how it will be exhibited in the future, so that the necessary steps can be
taken for climate control. If this information is not available, as is often
the case, or if there is too much uncertainty, the conservator will need to
apply a protection directly to the work, or in proximity to it, that is com-
patible with the principles already cited. It is hoped that these introductory
remarks and the presentations that follow make it clear that the author
does not believe in the existence of a miraculous substance or in a restora-
tion intervention that is capable of solving every kind of conservation
problem for panel paintings. Rather, it is possible to obtain good results by
General Criteria for
Conservation Intervention
316
Ciro Castelli
The Restoration of Panel Painting Supports
Some Case Histories
applying a series of interventions, whether they be in the form of treat-
ments or of preventive conservation eorts.
The primary goal in restoring panel paintings is to renew the func-
tionality of the structural support and to improve stability (with resulting
benets for the preparatory and paint layers) while adopting methods with
minimal invasiveness. The following examples of works restored during
the past few years in Florence will better clarify these concepts.
The rst example is The Coronation of the Virgin, an altarpiece painted by
Domenico Beccafumi in 1540 (Fig. 1). The painting, which comes from the
Church of the Santo Spirito in Siena, was executed for the Camaldolite
monastery of Ognissanti, outside of Porta Romana in Siena. After the
monasterys abolishment, the panel painting was moved to the Accademia
in Florence and exhibited until 1810. In 1832 Romanelli recorded it in the
sacristy of the Church of the Santo Spirito following its replacement with
the Annunciation by Girolamo del Pacchia. The work was nally placed
over the third altar on the right side of the church.
The Coronation of the Virgin
by Domenico Beccafumi
317 Tnt Rtsrorari o or Pati Pai ri o Surrorrs . Soxt Cas t Hi srori ts
Fi gure 1
Domenico Beccafumi, The Coronation of the
Virgin, 1540. Oil on panel, 310 187 cm.
Church of the Santo Spirito, Siena. Front of
the panel before restoration.
Diagnostic studies
To arrive at a precise understanding of the paintings support, it is impor-
tant to analyze the consistency of the wood and to establish with certainty
the various construction phases. Radiography and infrared reectography
(IR) are deemed useful tools for studying these aspects of panel paintings.
In general, radiography is the most eective analytical technique for exam-
ining the construction of the support and for identifying the state of
preservation of the wooden material (Fig. 2). For painted works, however,
the data this type of analysis can provide about the structural condition of
the wood bers are related to the thickness of the preparation and to the
presence of pigments that are particularly opaque to X rays. The X radio-
graph of The Coronation reveals wormholes in the support, as well as their
displacement close to the surface. This study also provided information on
the characteristics of the preparation, which showed up as slightly denser
in X rays of the lower section, a possible indication of a greater thickness
and dierent applications of the ground. Above all, the study revealed the
two-phase construction of the support. Naturally, this study was compared
with visual observations, an evaluation of resistance to touch on the back
of the support, and an assessment of the weight of the work in relation
to the type of wood. IR also proved useful for studying the support, as it
showed the preparatory image to be continuous between the upper and
lower sections. Thus, even if the preparation had been applied at dierent
times, the painting was conceived all at once.
Photographic documentation with diuse and raking light
revealed the state of preservation of the preparatory and paint layers.
The same techniques allowed documentation of the structural condition
of the support and the treatment carried out in the 1950s.
Construction technique
The painting, executed in oil on wood, measures 310 187 cm; it is
arched in the upper section. There was no cloth present as an isolation
layer between the wood and the preparatory layers. By 1540 such a
318 Cas t e l l i
Fi gure 2
Domenico Beccafumi, The Coronation of the
Virgin. Radiography of the lap join.
characteristic isolation layer had fallen into disuse, as the construction
technique for the preparatory layers no longer required the presence of
the cloth as a buer between the movement of the wood and the prepara-
tion.
1
The support is made of poplarmore precisely, white poplar
(Populus alba L.)and is formed of two distinct sections: an addition was
made to the already existing support before the application of the paint
layers. Thus, the support consists of two sections united with a 13 cm
wide lap join. The connection is reinforced with glue, as well as with nails
that are driven in from both the front and back and bent under the prepa-
ration (Fig. 3). The upper section comprises ve vertically oriented
planks.
2
The tree ring pattern is subradial, the quality is good, and the
presence of knots is rare. The lower section consists of six planks (also
oriented vertically)
3
of the same type of wood, with a medium tangential
cut. The planks of the entire panel, according to the customary technique
noted in the eld of Italian panel paintings, are arranged with the internal
side facing the preparatory layer; they are butt-joined and glued together
with lime casein.
The variations in width of the planks in the two sections and the
method by which they are joined give the impression that the two sections
of the support may have been built at dierent times. It is certain that the
extension was made before the paint was applied, because the painting pre-
sents a single pictorial composition, as revealed by visual and IR readings
and chemical analysis of the pigments. Conversely, the preparation was
carried out at two dierent times. This last piece of information, as already
mentioned, is conrmed by radiographic studies that showed a greater
density of the lower part of the painting, caused by the greater thickness
of the gesso layer. Last, the ground and paint layers of the lower part are
in better condition than those of the upper part. This is also true for the
wooden support, whose condition can be attributed to the use of better-
quality wood, which was almost certainly obtained from a dierent tree
with denser bers and greater resistance to attack by wood-boring insects.
A shaped frame (which is not the original) was placed along the perimeter
319 Tnt Rtsrorari o or Pati Pai ri o Surrorrs . Soxt Cas t Hi srori ts
Fi gure 3
Domenico Beccafumi, The Coronation of the
Virgin. Nail in the lap join of the painted
surface.
on the surface, covering 7 cm of the original paint. The frame was held in
place with screws, inserted from behind, that passed through the planks.
Apparently the back of the support had originally been sustained
and controlled by three crossbars, each attached to the painting by ve
small wooden brackets that were fastened with glue and with nails driven
in from the back and bent over on the front of the support. Both the nails
that connect the lap join and those used to attach the wooden brackets to
the support are simply bent, hammered into the wood, and covered by the
gesso preparation.
4
State of preservation
Upon the paintings arrival in the laboratory, large-scale lifting of the
preparation and paint layers was observed; this damage followed the grain
of the wood in the main section of the panel (Fig. 4). The failure of the
horizontal join on the back was also caused by the loss of extensive sec-
tions of worm-eaten wood that rendered several nails (those reinforcing
the connection between the two sections) isolated and useless. On the
front side of the area that corresponded to the join, a fracture aected
the preparatory and paint layers. In general, various glued joins between
the planks of the main support had opened. The stability of the paint layer
was good in the lower section; cracks were noted exclusively along the
joins. The state of preservation of the support appeared considerably
degraded overall from diuse attack by wood-boring insects that left the
wooden material extremely fragile and weak in some areas. These condi-
tions were worse in proximity to the vertical joins and the horizontal join
between the two sections; the greater degradation there can be attributed
both to the presence of protein substances from the glue and to the sap-
wood in the edge of each plank. The bottom section of the support did
not show the harmful eects of the wood-boring insects. Its damage prob-
lems consist essentially of gaps in the joins caused by a greater contraction
of these planks with respect to those in the upper part, and of a slight con-
vex curvature of the surface. All the wood used for the previous restora-
tion, particularly for the crossbars, had been extensively attacked by
wood-boring insects, leaving the wood extremely weak.
320 Cas t e l l i
Fi gure 4
Domenico Beccafumi, The Coronation of the
Virgin. Lifting of the paint.
Previous restorations
The restoration work done at the beginning of this century was carried
out with an invasive technique and with materials that were harmful to
the preservation of the wood (Fig. 5). This method created conditions
favorable to infestation by wood-boring insects and produced tensions
within the structure, causing the deterioration of both the support and
the painting. In previous restorations, the original crossbars had been
removed, and the missing areas had been reconstructed with a paste of
hide glue and sawdust.
5
Also, seven crossbars made of cypress and plane-
tree wood with an upside-down T cross section had been mounted to the
support with large, notched wooden blocks.
6
These elements had been fas-
tened with hide glue and large screws. Poplar strips had been attached
along the entire perimeter with glue, screws, and several nails hammered
in from the front. This interventionextremely invasive for the quantity
of wood added and for the method and materials usedalso made exten-
sive planing of the panel surface necessary. This planing facilitated the
exchange of moisture between the environment and the wood, a process
that, in turn, favored a tendency toward deformation of the planks, which
nevertheless was blocked by the interventions described above.
Another damaging intervention was the already cited application
of the frame to the painting; the operations necessary to adjust the frame
and hold it in place had produced twenty holes (each 6 mm in diameter) as
well as six deep tracks into the painted surface (Fig. 6).
7
Restoration proposal
Given the severely deteriorated state of preservation of the wooden sup-
port and the ground, the following program was outlined:
1. Consolidation of the most degraded parts of the wooden
material with acrylic resin (Paraloid B72).
321 Tnt Rtsrorari o or Pati Pai ri o Surrorrs . Soxt Cas t Hi srori ts
Fi gure 5
Domenico Beccafumi, The Coronation of the
Virgin. Back of the panel before restoration.
Fi gure 6
Domenico Beccafumi, The Coronation of the
Virgin. Marks in the painted surface caused by
adjustment of the frame during previous
restoration.
2. Reinforcement of the addition by the insertion of several
rectangular wooden pieces to hold the two parts of the
join together.
3. Construction of two temporary crossbars to hold the painting
during the removal of the existing crossbars.
4. Gradual removal of the existing crossbars.
5. Repair and correction of the separated edges by the cutting of
tracks with a V-shaped section.
6. Leveling of the painted surface along the edges of the indi-
vidual planks.
7. Exact tting and placement of the wedge-shaped inserts, to be
made of old wood (of the same type as the support), into the
specially prepared V-shaped tracks.
8. Construction of a laminated oak framework that has a load-
bearing function and also controls the deformation of the
planks that make up the support.
9. Development of a plan for the microclimate control of the
back of the panel.
Restoration interventions
Fumigation
The painting was fumigated in a gas chamber without vacuum, and then
protected on the back with Permethrin, a fumigant that remains active.
Consolidation of the wooden material
Before the removal of the crossbars and the wooden blocks of the previ-
ous restoration, the join between the two sections of the support, which
was in danger of separating, was reinforced. The technique used for this
operation consisted of placing twelve rectangular inserts, made of the
same wood as the support, on edge in the same grain orientation as the
bers of each plank, positioned across the horizontal junction. These
inserts were distributed in the grain direction of the planks, penetrating
the thickness of the support to within 5 mm of the painted surface. Thus
the inserts reunited the two elements of the lap join.
One of the more problematic aspects of this restoration was the
need to regain sucient strength in the areas of the wood that were
degraded by biological attack. The choice of consolidant was proposed in
consideration of the uncertainties consolidants had generated in the past
in the Florence laboratory; particular concerns were the ecacy of con-
solidants and their possibly negative eects over time. These concerns are
tied to the stability of the product, possible color alterations, and nonuni-
form penetration into the wood (so that dierent areas of the wood are
conditioned to respond dierently to variations in RH). In this case, how-
ever, it was decided to use a 512% solution of acrylic resin (Paraloid B72)
in lacquer thinner applied by brush until a sucient consistency was
reached. Before this operation began, all the hide glue and sawdust llings
were removed from the support, and two temporary crossbars were made.
These crossbars were modeled to the curvature of the painted surface to
support the panel adequately and make it possible to work on the back.
The removal of the wooden blocks that held the existing cross-
bars followed; this procedure freed the entire back surface of the support
and prepared it for the initiation of the consolidation technique. Repair and
322 Cas t e l l i
reconstruction of the structure proceeded with the reintegration of the
missing parts of the support, in particular in the lap join. For this operation,
small blocks of old wood (of the same type as the support) were placed in
superimposed layers that intersected in width and length (Fig. 7). The use of
this method makes it possible to rmly bond the various wooden elements
of the reintegration and, in addition, favors increased stability by reducing
to a minimum the possible deformation of the added material.
The repair and rejoining of the separated joins and cracks were
carried out by the cutting of triangular tracks into the support. These
tracks conformed approximately in width and depth to the extent of the
degradation in the areas of the join edges.
8
The tracks were cut by hand with chisels (the traditional and
eective method to rectify gradually the degraded condition at the edges
of each plank), but wherever the consistency of the wood was good
and the split was straight, an extremely narrow, cone-shaped router bit
(5 mm maximum diameter) was used in order to remove as little original
wood as possible.
Successful experiments had already been performed with this bit,
made expressly for our laboratory, on samples that simulated V-shaped
openings in panel paintings. As usual, in the preparation of the wedge for
gluing, the surface levels along the joins and splits were aligned with the
help of wooden levers. These wooden levers bridged the edges of the frac-
tures and were adjusted with screws and wooden blocks. This step was fol-
lowed by the tting of the wedges, which were made from old wood (the
same type as the support). Care was taken to ensure that the positioning
of the grain was consistent with the grain at the edges of the opening.
During this work phase, the undulation of the painted surface, caused by
the curvature present in the two central planks, was slightly corrected, so
that part of this deformation was distributed over the entire width of the
painting. This operation has been shown to be useful in reducing the
deformations visible near the joins and in improving painting readability.
9
The correction of the edges produced an average curvature of 9 mm over
the entire painted surface. It is not useful to plot deformation measure-
ments without considering the ambient RH, because the wood is in con-
stant equilibrium with the surrounding microclimate and consequently
323 Tnt Rtsrorari o or Pati Pai ri o Surrorrs . Soxt Cas t Hi srori ts
Fi gure 7
Domenico Beccafumi, The Coronation of the
Virgin. The integration of missing parts in the
support with blocks of old wood.
continually modies its warp. The RH considered suitable to ensure the
stability and uniformly at surface of the wooden support varies between
55% and 60%. A polyvinyl acetate (PVA) emulsion was used to adhere the
wedges, since its strength and moderate elasticity enable it to adapt better
than other glues to the conservation needs of wood.
Support control system
After repair and reconnection of the joins, reconstruction of the missing
wood parts, and reinforcement of the junction of the lap join, the support
appeared quite solid. The only remaining phase was the construction and
mounting of a crossbar system to control and reinforce this particular
construction.
The author selected a system that could simultaneously respond
to the expansion and contraction and, in addition, serve as a sound rein-
forcement, assuming the role of a true load-bearing structure. A perimeter
strainer, or framework, was built with crossbars made of laminated oak
from Slovenia. The crossbars had the same 9 mm curvature as the support,
so that the strainer would conform to the shape of the panel. The frame-
work was attached to the support without leveling of the back surface.
Instead, small wooden spacers were inserted at the attachment points
where the contact between the two parts was not perfect. A special mecha-
nism to unite the two parts allows for potential expansion and contraction
of the support and regulates possible warping of the planks. This mecha-
nism consists of a brass shoe in the form of a closed U-channel section,
held to the back of the support with a single screw (Fig. 8). Inside this
U-channel section glides a nylon slide with a bolt at the center. The bolt
passes into the framework through a brass sleeve, in which there is a
spring regulated by a nut.
10
The invasiveness of this mechanism to the painted support is
limited to a single screw for each element. The presence of the spring
between the support and the framework facilitates the regulation of stress
and possible slippage between the two parts, as well as reduces tension.
Designed to respond to problems of tension and deformation that may
appear over time, the mechanism is extremely simple and does not require
any intervention to the support (Fig. 9).
The selection of a means of deformation control with a frame-
work system goes against the concept of the traditional crossbars, which
act as a load on the support. Instead, the framework is a load-bearing
structure to which the painting is anchored by means of attachments freed
from mechanical tensions. In addition, if the RH of the exhibiting environ-
ment is uncertain, the framework makes it possible to enclose the back of
324 Cas t e l l i
Fi gure 8
Example of the mechanism that attaches the
framework to the painting support.
the panel easily. This enclosure creates a volume of air that slows climatic
exchanges with the environment, thus facilitating stabilization. At the end
of the intervention, the support is protected actively by a brush applica-
tion based on Permethrin. The protection of the reverse is completed with
a mixture of beeswax (60%), paran (30%), and rosin (10%) applied with a
spatula to form a surface lm.
11
This treatment can prevent new infesta-
tions and slow the rate of air exchange with the environment.
From 1987 to 1988, the author was involved in the restoration of three
panel paintings from the Flemish school. These works by the painter
Herri met de Bles (nicknamed Il Civetta) came from the Museum of
Capodimonte in Naples. Two of the paintings presented conservation
problems that also involved the wooden support. The following para-
graphs describe one of these works with regard to its construction char-
acteristics, its particular conservation problems, and the restoration
intervention to which the work had previously been subjected (Fig. 10).
Jesus and Saint Peter on the
Water by Herri met de Bles
325 Tnt Rtsrorari o or Pati Pai ri o Surrorrs . Soxt Cas t Hi srori ts
Fi gure 9
Domenico Beccafumi, The Coronation of the
Virgin. Back of the panel after attachment of
the framework for support and deformation
control.
Fi gure 10
Herri met de Bles, Jesus and Saint Peter on the
Water. Oil on panel, 26 37 cm. Museum of
Capodimonte, Naples. The state of preserva-
tion of the panel before restoration.
The restoration intervention that can address such conditions eectively
requires special solutions in methodology and technique.
Diagnostic studies
Full-scale and detailed photographic documentation was carried out with
diuse and raking light. Raking light photography revealed the type and
quantity of the lifting paint on the painted surface, as well as the supports
deformation, especially in areas aected by cracks. Low-magnication
observation was all that was required to identify the wood species, as
the type of wood grain, the color, and the characteristic sheen of the
parenchymal rays left no doubt about its identication as oak. While
the RH was kept constant, relief drawings were made on graph paper to
determine whether the curvature varied after the cracks were rejoined.
Construction technique
This small oil painting on panel consists of a single board of oak (Quercus
peduncolata or Q. sessiliora). The board has a straight, horizontal grain,
with the tree rings positioned subradially. No knots or defects were noted
in the support. Cloth was not used for the preparatory layers. It was clear
that crossbars had never been used, both because of the paintings small
size (26 37 cm and currently 34 mm thick) and because of the custom-
ary way supports were made in the Low Countries. In fact, for the great
majority of these panel paintings, deformation is controlled and the wood
bers are supported horizontally and longitudinally simply by means of
the frame. The frame had a channel routed in its thickness that made it
possible to enclose the painting around the perimeter without restricting
eventual expansion and contraction. A few exceptions to this rule employ
reinforcement crossbars on the back.
State of preservation
The painting presented diuse lifting of paint along the grain of the wood,
as well as warping of the painted surface, which could be seen in three
pronounced curves. At the edges of these deformations were two cracks
that followed the grain, aecting the entire width. Although oak character-
istically has a mechanical strength, durability, and resistance to wood-
boring insects, the general conservation conditions were decidedly
precarious. The wood, especially along the borders, was eroded and fri-
able. The diuse attack of wood-boring insects had produced many cavi-
ties, some of which had a diameter equal to half the thickness of the
support. The weakened mechanical resistance of the support bers was
aggravated by a crossbar system that presented two drawbacks: it was
extremely rigid in comparison to the size of the painting, and it func-
tioned as a brace at a distance of only about 5 mm from the plane of the
support. The planing of the back of the support contributed to the deterio-
ration of the panel by causing the loss of the surface skin of the wood,
so that a more rapid exchange of moisture between the support and the
environment was encouraged.
Previous restoration
The last restoration of this painting occurred in the early 1950s. At that
time, consolidation of the ground and paint layers, cleaning, lling of
326 Cas t e l l i
losses, and retouching were done. The back of the support had been
planed down, a procedure that removed a small amount of wood. Three
mobile crossbars were attached to the panel with poplar blocks, positioned
in line with the grain of the panel, and glued in place (Fig. 11). The cross-
bars were circular-section aluminum rods that passed through holes made
in the blocks attached to the support.
Restoration proposal
The following solutions were identied: removal of existing tension in the
support; consolidation of the ground and paint layers; repair of the worm-
holes that had weakened the wood; and development of a sound support
and control system for the panel. All of these operations had to take place
with minimal invasiveness to the supportin accordance with a philosophy
that is increasingly valued in the Florence laboratory. In this particular case,
it is apparentgiven the small size of the supportthat excessive use of
wooden material and glue could potentially damage the painting over time.
Restoration interventions
With the painted surface protected by Japanese rice paper and rabbit-skin
glue, the consolidation of the paint layer was carried out by the vacuum
technique with the same type of glue in a dierent concentration.
12
Two
temporary crossbars were constructed to hold the painting in its current
deformed state. A gouge was used to remove the supports that held the
crossbars added during the previous restoration, thus liberating the support.
While the temporary crossbars held the support orthogonal to the grain,
the cracks were repaired by small V-shaped tracks opened with the tradi-
tional chisel method. With this operation, the two faces of the cracks were
aligned and prepared for the wedges, and the painted surface was leveled.
This initial phase was essential in giving the disjointed and deformed front
faces a uniformly at surface. It also made it possible to rotate slightly the
disconnected edges of the cracks, while still preventing the back edges of
the cracks from touching. In this way the panel took on an uninterrupted
surface in correspondence with the cracks. To arrive at this solution, the
painting was inserted into a special cagelike structure in which the correc-
tion of the warp and the alignment of the edges was begun (Fig. 12). This
structure, built especially for this project, makes it possible to enclose the
painting at the bottom, top, front, and back. The author and others were
able to work on the edges of the opened cracks and adjust the levels of
the painted surface by means of screws (the heads of which are protected
by wooden caps) that can slide inside the vertical slats of the cage struc-
ture. With the aid of this system, the temporary crossbars were removed
and the prole of the painting corrected. After this procedure, the wedges
made from old oak were tted, in correspondence with the orientation
of the grain, and a PVA emulsion was used to hold them into place. This
operation was repeated with the other crack.
The holes caused by the wood-boring insectsthe problem that
posed the greatest threat to the structural soundness of the supportwere
rebuilt with inserts made from the same type of wood as the support.
Triangular or rectangular insertsdepending on the shape and depth of
the holeswere held in place with PVA emulsion.
To restore solidity, control, and protection to the edges of the
painting, a perimeter framework was made with the same curvature as the
327 Tnt Rtsrorari o or Pati Pai ri o Surrorrs . Soxt Cas t Hi srori ts
Fi gure 11
Herri met de Bles, Jesus and Saint Peter on the
Water. Back of the panel upon its arrival in the
laboratory.
back of the support, and a central crossbar was installed. The framework,
made of chestnut, was anchored to the support with nine springs. The
springs were attached to the framework on one end and to the painting on
the other end with an equal number of small blocks (9 9 4 mm thick).
These blocks, made of the same type of wood as the support, were cre-
ated with a hole to house the end of the spring. They were attached in the
direction of the grain and glued in place. The blocks were placed inside
the framework, and a nearly 2 mm space perpendicular to the grain was
left to allow for possible expansion of the panel. Elastic control of the
warping is provided by the springs, primarily by those positioned in con-
formity with the central axis (Fig. 13). Thanks to its solid and stable
construction, the framework protects the edges and provides a secure sup-
port for the panel. This type of construction to control movement of the
support does not put any weight on the panel, as do traditional crossbars.
Instead, the panel is supported by anchor points distributed over the sur-
face. Because of the reduced size of the frameworkcorresponding to
the small size of the artworkand the small blocks glued to the support,
which allow the springs to connect the framework to the panel, it is pos-
sible to reduce substantially the invasiveness of the intervention (Fig. 14).
In future conservation eorts, it may be possible to adjust the ten-
sion in the springs without tampering with the anchor points of the sup-
port. This use of springs to control deformation is best applied to supports
that consist of a single board, as independent deformation of other boards
is not a factor. This structure can be closed on the back; the backboard cre-
ates a volume of air that functions as a buer, slowing RH variations. The
wood used can be of the same type as the painting support. Such a device,
already described in the preceding intervention, slows climatic exchanges
between the back of the support and the environment. Because of the
small size of this painting, the back enclosure also provides increased sta-
bility to the support, augmenting the wood mass by lling up the frame-
works two cavities. To make this possible, the wood put inside the
framework must be oriented with the grain of the panel and placed so
that it is completely independent of the panel. Naturally, the restoration
of the support also protects the back from wood-boring insects.
328 Cas t e l l i
Fi gure 12
Temporary jig for adjusting the surface level.
Fi gure 13
Spring attaching the framework to the
support.
This panel painting (495 285 cm) is executed in oil on panel and was
painted around 154748 for the Dini Chapel in the Church of Santa
Croce in Florence.
Diagnostic studies
The analyses done in preparation for the restoration of the wooden sup-
port consisted of measuring the moisture content in the wood. This was
accomplished with the aid of probes attached to the planks (other probes
measured the RH in the surrounding environment). All the probes were
connected to a computer. The aims of this survey were to establish the
relationship between the wood and the environment and to determine the
importance of the reaction of the support to variations in these values.
For this inquiry, a gauge was used that made it possible to obtain the val-
ues of the horizontal expansion with a centesimal scale.
13
For the curva-
ture, the control was simply a reference plane. Traditional photographic
documentation of the condition of each plank was carried out. Particular
attention was given to the parts of the painted surface that were aected
by such problems as lifting paint, detachment of the original inserts (origi-
nally placed to repair large knots), cracks, and defects in the wood. Finally,
several interesting details of the original construction technique were doc-
umented. Identication of the wood species was made with macroscopic
and microscopic studies.
Construction technique
The painting, which is arched at the top (Fig. 15), was constructed without
cloth between the wood and the preparatory layers; strips of cloth are not
even found along the joins. The support consists of six planks of poplar
more precisely, white poplar (Populus alba L.). The planks are of average
size with solid and relatively straight grain.
14
The exception is the second
plank from the left (seen from the back), which presents a curvilinear
grain. The planks are of a medial cut, and the bers are generally arranged
subradially at the edges of each plank, becoming tangential at the center.
Two radially cut planks that contain the pith are an exception to this char-
acteristic. Because the plank is wavy along the length, the surface level
changes from one plank face to the otherhigh on one side of the joint
and then high on the other side. In addition, a considerable number of
large knots, which have had a negative impact on the preparatory layers,
The Deposition from the Cross
by Francesco Salviati
329 Tnt Rtsrorari o or Pati Pai ri o Surrorrs . Soxt Cas t Hi srori ts
Fi gure 14
Herri met de Bles, Jesus and Saint Peter on the
Water. Back of the panel, with the framework
for support and deformation control in place.
were noted. Because of these defects, many planks were repaired during
the original construction with the application of plugs of wood (also of
white poplar) held in place with hide glue and nails. The assembly of the
planks was achieved by butt-joining and accurate planing of the faces to be
united. Diagonal scratches were also made, to improve the bond of the
glue. Housings were carved inside the thickness of the planks, in proxim-
ity to the joins, in which oating tenons were inserted (Fig. 16).
15
These
elements of joining between the planks, regularly spaced in height on the
painting, are made of walnut and have a rectangular shape. They were
inserted without glue into the housings with the grain perpendicular to
that of the support, and held by dowels that pass through the thickness of
the planks. Three r crossbars that tapered in length were mounted on the
back and inserted into dovetailed tracks cut into about one-third of the
thickness of the support. The panel is relatively thin for its size and under-
goes only light restraint from the crossbars. It was discovered that in real-
330 Cas t e l l i
Fi gure 15
Francesco Salviati, The Deposition from the
Cross, 154748. Oil on panel, 495 285 cm.
Church of Santa Croce, Florence. View of the
front of the panel shows the separated planks
before wood restoration.
Fi gure 16
Francesco Salviati, The Deposition from the
Cross. Side view of the plank, showing parts of
the original oating tenon and the diagonal
scores made to improve the hold of the glue.
ity, the large carved and gilded frame that held the panel not only served
an aesthetic purpose but had a structural function as well.
State of preservation
The painting had been immersed during the ood of 1966 in Florence.
During this time, the water, which was full of various materials, covered
four-fths of the altarpiece for approximately eighteen hours. In the
Museum of Santa Croce, the panel was immediately protected with tissue
papers of various sizes that were made to adhere to the surface with
acrylic resin (Paraloid B72). Next the painting was moved to the limonaia
in the Boboli Gardens, where the humidity of the environment was pur-
posely kept at 9095% to protect ooded artwork. The negative eects of
the immersion on the planks are well known; the initial reaction was an
expansion of the surface, after which the preparation and paint layer were
loosened by water. During the next phase, the various materials within the
painting dried at dierent rates, causing detachments and the overlapping
of the panels preparation and paint layers during shrinkage.
The initial papering done on site was repeated several times while
the painting was sheltered at the limonaia. The rst intervention on the
wooden structure was the mechanical removal of the original crossbars.
16
Before this operation, the painting was laid at and faceup on a wooden
structure that made it possible to work from below. Rectangular wooden
blocks applied to the back bridged the joins and attached to the planks by
two screws on both ends of the blocks. This served to reinforce the joins
in order to hold together the planks that made up the painting. When it
was brought to the laboratory at the Fortezza da Basso in June 1967, the
panel was already separated at the joins, with the exception of a small part
in the upper right of the last plank. The entire structure remained united
only by the oating tenons that were left in the housings and held by the
dowels placed at the ends of the tenons. There were convex warps on the
painted surfaces of the planks, and two of the lower planks had become
concave and twisted in the longitudinal direction. The preparatory layers
were extremely deteriorated and unstable. The plugs placed to repair the
knots exhibited their own deformations: they had detached from their
housings and marked through onto the painted surface (Fig. 17). The par-
ticular characteristics of the deformations in this paintingrelated to the
331 Tnt Rtsrorari o or Pati Pai ri o Surrorrs . Soxt Cas t Hi srori ts
Fi gure 17
Francesco Salviati, The Deposition from the
Cross. An original plug present in the painted
surface; it had come unglued during the
panels immersion in oodwaters during the
ood of 1966 in Florence.
direction of the grainmanifested themselves in two planks that were
simultaneously concave at the bottom and convex at the top.
Restoration interventions
The painting was followed through its stabilization phases for many years.
During this time, thorough consideration was given to possible working
solutions for consolidating the paint layer, guaranteeing sucient stability,
and restoring the lost unity of the entire work by the application of a sup-
port and control structure for the planks.
The traditional intervention technique often used in such cases
involves destroying the wood and transferring the paint layer to another
support, incurring all the changes and risks connected to this type of opera-
tion. In this case, however, the conservator followed an intervention that
would respond to the criterion of greatest possible respect for the original
components and that would, in addition, allow the possibility for a later
intervention. After evaluating the results on the consolidation of the paint
layer, the author and coworkers designed and carried out the restoration
of the support. This plan required an intervention on each individual
plank to repair the original detached and disjointed plugs, upon removal
from the painted support, by cutting the anchoring nails. Next came the
addition of wooden blocks into the housings of the plugs in a parquet
fashion (Fig. 18).
17
This procedure was followed by a reduction of the
thickness of the plugs to facilitate reestablishment of the level between
the blocks and the surface of the painting.
Each section was adhered again to its own place, so that the
proper level between the edges of the paint layer was re-created (Fig. 19).
The wedge technique was used to close the cracks at the edges of the
planks. Finally, the author inserted into the original tracks of the crossbars
a double layer of small pieces of white poplar, placed in the same grain
direction as the support.
Filling the original tracks with a double-block system superim-
posed widthwise and, especially, lengthwise, responded to the need to
improve the adhesion between the added parts and the original panel, par-
ticularly in the areas where there is an end-grain join (Fig. 20).
18
Next the
planks were rejoined through a slight correction of the edges, so that a
solid union could be obtained by means of wedges.
19
For this operation the
332 Cas t e l l i
Fi gure 18, bel ow
Francesco Salviati, The Deposition from the
Cross. Repair of the housings and plugs.
Fi gure 19, bel ow ri ght
Francesco Salviati, The Deposition from the
Cross. Treatment to re-create the proper level
between the edges of the paint layer.
author aligned the planks according to a regular curvature that followed
the individual deformations of the planks and that took into account both
the visual unity of the work and its structural needs. This operation was of
critical importance and required a lengthy time for study and the prepara-
tion of dierent simulations to help determine the proper equilibrium
between the deformations of the individual planks and the general curva-
ture that derives from them. When the correct equilibrium was reached
between the curvature and a good visual unity of the work (with its place-
ment in the original frame considered), two temporary crossbars were
made that served as a reference during the nal assembly phase.
Another diculty was the need to bring together at an equal level
the painted edges between the individual deformed planks. Through
observation of the defects that emerged in this painting, it was possible to
conrm that the deformations that appeared are not casual ones but
clearly respond to the composition and direction of the grain and, more
particularly, to the arrangement of the tree rings in the planks. This infor-
mation also conrms that the planar stability of a painting hinges on a
careful selection of wood at the time of construction.
The author and coworkers then began work on rejoining the
panel from the two central planks, starting from the center and moving
toward the outside, using the already described crossbars as a reference,
and proceeding gradually both with the leveling of the painted surface
and with the gluing of the wedges (Fig. 21).
20
The remaining parts of the
painting were reassembled with this same method (Fig. 22).
After this phase, a crossbar system was built that was identical
to the original in both the kind of insertion used for the crossbars and
the mode of function. This system, which appropriately limits the expan-
sion and deformation over the entire surface, seemed the most suitable
for the state of conservation of the pictorial surface of this particular
work. Movement is controlled by the friction encountered by the crossbar
within the tapered, trapezoidal shape of the track in the support (panel).
Conversely, the elasticity of appropriately sized crossbars controls defor-
mation. To function appropriately, the crossbars were made with a curved
prole, part of a circle that follows the curvature of the support. This
333 Tnt Rtsrorari o or Pati Pai ri o Surrorrs . Soxt Cas t Hi srori ts
Fi gure 20
Francesco Salviati, The Deposition from the
Cross. Wood-block system used to ll in the
tracks of the original crossbars.
operation was accomplished with the double-block system used in the
tracks of the original crossbars. The original tracks were widened into a
new track, which maintained the trapezoidal section of the original. The
new crossbars were made out of laminated oak assembled on a precon-
structed negative form; they were fashioned with the same curvature as
the track in the panel (Fig. 23). The crossbar dimensions were determined
by the large size of the entire work and by the designs likely stability over
time. The frame will regain its important structural function, completing
the support of the painting as it did originally.
The back of the painting was next protected with a thin coating of
beeswax, paran, and rosin spread on with a spatula. Isolation from the
environment was further guaranteed by placement of the work inside the
original frame and installation of a backboard. This feature will make it
possible to improve the climatic factors in contact with the support.
One additional example completes the presentation of the various operat-
ing methods that can be applied to the restoration of painting supports.
Changes are not introduced for their own sake but as a study of solutions
to the problems of individual works. This exible attitude is vital and is
based on the understanding that every intervention that changeseven a
littlethe original construction, if not required by the state of preserva-
tion of the work, is to be considered a loss of evidence and cultural patri-
mony. The restoration presented briey here was recently concluded on
the triptych, The Annunciation, by Lorenzo Monaco (Fig. 24). This work
was executed in the period from 1424 to 1425 for the Bartolini Salimbeni
family; it came from the Church of Santa Trinit in Florence, where it
had been installed in the fourth chapel on the right side of the nave. As
with the preceding examples, the particular choices that governed the
The Annunciation by
Lorenzo Monaco
334 Cas t e l l i
Fi gure 21, ri ght
Francesco Salviati, The Deposition from the
Cross. Reassembly of the planks.
Fi gure 22, f ar ri ght
Francesco Salviati, The Deposition from the
Cross. Reassembly. Seen from the front.
Fi gure 23
Francesco Salviati, The Deposition from the
Cross. End view of the inserted crossbar.
intervention follow the rule of respect for the original construction while
also providing an appropriate functionality to the support and conferring
the best stability to the paint and preparatory layers.
Diagnostic studies
With regard to the wooden construction of this painting, the current
state of preservation and the deformations present in the panel were
documented. The construction technique was analyzed and the wood
species identied.
Construction technique
The painted panel (265 236 cm), with three cusps in the upper part, con-
sists of a support of seven planks made of white poplar oriented with the
grain in a vertical direction and butt-joined with lime casein glue (Fig. 25).
Inside the joins, at the center of the thickness of the planks, are wooden
dowels that connect the planks. Most of the planks were obtained with
medial cuts; only the central plank was radially cut. The technique for the
painting preparation uses a cloth and a thick layer of gesso and glue; the
painting medium is egg tempera. On the upper part of the panel front,
there are also cusps superimposed on the main plank, with carved ogival
framing elements that contain, inside tondi, the gures of God the Father
and the prophets. Below, there is a small predella with inscriptions. The
work itself rests on a larger, stepped predella, where scenes from the
Nativity are depicted.
The support is reinforced on the back by three crossbars of
poplar, placed at right angles to the panel planks and xed with nails,
which were driven in from the front prior to the preparation and then bent
over on the back of the crossbars. The small predella at the lower edge of
the painting is made of a board placed at a right angle to the grain of the
support and fastened with nails.
335 Tnt Rtsrorari o or Pati Pai ri o Surrorrs . Soxt Cas t Hi srori ts
Fi gure 24, bel ow
Lorenzo Monaco, The Annunciation, 142425.
Egg tempera on panel, 265 236 cm. Church
of Santa Trinit, Florence. Photographed in
raking light.
Fi gure 25, bel ow ri ght
Lorenzo Monaco, The Annunciation. Back of
the panel photographed in raking light.
State of preservation
The planks that make up the support show slight convex warping on the
painted surface. This phenomenon is greater on the outside planks, and
on the painted surface, a pronounced misalignment is noted at the open
splits. The central plank, from a radial cut, shows separation of the tree
rings in correspondence with the pith. The phenomenon of ring separa-
tion (the onion eect) is typical for chestnut but rare in poplar (Fig. 26).
Finally, the joins were open, and cracks were noted in the bottom part
of the support.
Previous restorations
The last restoration carried out on the painting dates from the 1950s. On
that occasion, the paint layer was cleaned, and the gesso that had been
applied on the gilding of the architectural part was removed. To restrain
the planks that had separated, buttery inserts were inset across the joins.
Some of the planks of the support were warped, so that a double convex
deformation was formed across the painted surface. To lighten the tension
in the planks, the crossbars were reduced in thickness. The upper one
was thinned though not removed from its housing, while the central
one was removed by the cutting of its original nails; it was then thinned
and put back into place with normal screws. Finally, in the case of the bot-
tom crossbar, it could be seen that the nails had been straightened and the
crossbar slipped o, reduced in thickness, and reattached with the same
elements, since the heads of the nails could be reached by removing the
smaller predella.
Restoration proposal
The plan for the intervention was established after careful evaluation of
the condition of the painted surface, the consistency of the wood, the
uneven surface alignment at the splits, and the hold of the crossbars.
The removal of buttery inserts was justied in that their func-
tion is only partial, and, in fact, they are even harmful, since the grain is
placed in the direction opposite to that of the support, so that tension
points are created between the planks.
The following were planned: repairing the splits with the wedge
technique, adjusting the surface level at the splits, overhauling the cross-
bars, and protecting the back with an antiwoodworm product based
on Permethrin.
Restoration interventions
By use of an electric router attached to a pantograph template, the wal-
nut buttery inserts were removed. This made it possible to obtain the
perfect refacing of the cavities on both the edges and the bottom. In this
case the author felt that the slight vibration of the instrument would be
well tolerated by the support and by the entire, rather solid preparation-
paint layer. The cavities were corrected and lled with small elements
of old wood of the same type, and arranged in the same grain direction,
as the support.
The separated edges of the joins and cracks were repaired with
the traditional method: triangular-section blocks were custom-tted into
336 Cas t e l l i
Fi gure 26
Lorenzo Monaco, The Annunciation. A split
on the back of the panel; some ring separa-
tion is evident in the wood of the supports
central plank.
prepared tracks and glued in place with a PVA emulsion.
21
The degraded
part of the wood was eliminated by the correction of the edges, and the
level of the painted surface in those areas was realigned.
In this operation our intervention was limited to removing only
the parts aected by degradation, so that, as much as possible, the trian-
gular angle of the cut was retained. It was considered vitally important
that each wedge be positioned in such a way that the grain be parallel
to that of the panel and that the annual rings be arranged radially with
respect to the plane of the support. Such positioning is more compatible
with the wooden construction and, in case of dimensional changes, guar-
antees less deformation and, thus, a greater bond to the support. The
most dicult part of this operation was the repair of the onion eect seen
in the central plank (Fig. 27). The author and coworkers thus proceeded
with the removal of the wood aected by the phenomenon, following its
irregular disposition. The intervention was carried out in multiple steps,
by alternating opening, leveling, and gluing of blocks in many layers for
the reconstruction of the weak parts. Then wedges were inserted to join
the faces of the opening. This technique made it possible for the surface
to be perfectly adjusted: the multistep integration reduced the forces in
the wooden material, lending greater stability to the intervention. Even
the leveling of this area had to be done in dierent phases to obtain good
results without the application of especially strong force. The solution
adopted for the crossbars followed the concept used in the previously
described interventionthat is, to make modications only in the bonds
between the various components.
The method of reducing the rigidity of the crossbars is altogether
valid today when the panels planks are subjected to a warping stress that is
thought to be irreversible. Such situations generally are the cause of the for-
mation of cracks and instability of the paint and preparatory layers. These
phenomena can be caused by the aging of the material in relation to the
characteristics of the wood in terms of quality and positioning of the grain,
or by environmental factors that have aected the life of the painting.
Removal of the upper and lower crossbars proceeded with the
cutting of the tips of the nails that were hammered over onto the back.
This operation made it possible to observe that two sides of nails had
some space in the horizontal direction as a consequence of the yielding
between the parts (the walls of the wood and the exible metal) that
occurred over time. The next step consisted of widening the holes left by
the nails in the crossbars (by 3 mm) and threading (with a 4 mm pitch) the
uppermost centimeter or so of the nails that protruded from the support
for the entire thickness of the crossbars. The crossbars were then inserted
onto the protruding nails. Because the crossbars had been thinned on
the back to a thickness of a few millimeters, it was possible to reconnect
the crossbar to the support in a stable manner with a nut. For the central
plank, which was missing the original nails, a mechanism was used that
consisted of a brass stop plate with a rectangular slot inside, held onto the
lower face of the crossbar in contact with the panel with two screws and
epoxy resin; a bolt was free to move within the slot but was held in by its
head. A double-threaded brass bushing was inserted into the supportthe
external thread to anchor the bushing to the support, the internal thread
to receive the bolt. The bolt passes through a hole made into the crossbar
and attaches itself to the support. Movement is ensured by the head of the
337 Tnt Rtsrorari o or Pati Pai ri o Surrorrs . Soxt Cas t Hi srori ts
Fi gure 27
Lorenzo Monaco, The Annunciation. Ring sep-
aration in the wood on the back of the panel
before repair. Also visible is the reconstruc-
tion in the grain direction of the old buttery
inserts.
screw, which is held by the slot of the stop plate, and is aided by the inter-
position of two convex washers and one Teon washer (Fig. 28).
Choosing to remove the existing crossbars in this case is based on a
study of the equilibrium between the crossbars and the support planks, and
on the stability of the paint and preparatory layers. It did not seem appro-
priate to intervene with new crossbars, even if they would function better,
because of the risk of disturbing the existing equilibrium. The chosen inter-
ventionwhich was believed to be sucient to guarantee the solidity of
the structurepreserved what remained of the original crossbars, limiting
the intervention to reestablishing the integrity of the support with the
repair of the cracks. This operation interrupts the circulation of the micro-
climate through the openings present between the planks, guaranteeing
greater stability. Naturally, taking appropriate precautions for the paintings
exhibition in the church, preparing a microclimate analysis of the environ-
ment, and establishing the appropriate interventions are necessary.
Restoration of the works presented here was carried out at the Opicio
delle Pietre Dure e Laboratori di Restauro by the following individuals:
Dr. Marco Ciatti, director of the work; C. Castelli, M. Parri, A. Santacesaria,
R. Buda, P. Bracco, O. Ciappi, N. Bracci, T. Cianfanelli, and M. Rosa Sailer,
restorers for the wooden supports and painting; Dr. A. Aldrovandi,
reectography studies; Dr. A. Aldrovandi and O. Ciappi, radiography stud-
ies; C. Castelli, M. Parri, and A. Santacesaria, 35 mm photography; Sergio
Cipriani and Lamberto Cerretini, archive photos; A. Santacesaria, AutoCad
designs. I thank the many colleagues and friends who collaborated on the
work presented herein particular, Dr. Giorgio Bonsanti, superintendent
of the OPD, Dr. Cecilia Frosinini, and Dr. Marco Ciatti for their collabora-
tion in outlining this article. Further thanks to my friends Andrea Rothe
and George Bisacca for helping with the translation from Italian.
1 The use of cloth for the preparation of panel paintings is noted from 1138 (Croce di Sarzana)
until the early 1400s. This element reduces the eects of settlement and movement of the sup-
port wood, helping to preserve the preparatory and paint layers. A heavy layer of gesso and
glue several millimeters thick is added. This technique is described by Cennino Cennini in his
Libro dellarte (ca. 1437). After this period and throughout the 1400s, cloth or parchment strips
were applied on many panel paintings in correspondence with joints, nailheads, and imperfec-
tions in the wood.
Notes
Acknowledgments
338 Cas t e l l i
Fi gure 28
Model of the anchoring system used for the
central and lower crossbars.
2 The widths of the planks that compose the upper support measure as follows, starting from
the left: 21, 46.5, 56, 49, and 13.5 cm.
3 The widths of the planks of the supports lower section measure as follows, starting from the
left: 15, 38.5, 31, 40, 33, and 28.5 cm.
4 Of the several ways to guarantee that the heads or tips of nails do not contact the preparatory
layers, the most suitable method recreates a uniform support surface for the preparatory layers
by recessing the nail a few millimeters into the thickness of the support and then covering it
with a wooden plug. This method is most prevalent in the oldest works. In less careful prepa-
rations, pieces of cloth or parchment were applied with the aim of isolating the metal.
5 A mixture of paste, made with hide glue and sawdust, is frequently found in restorations of
painted wooden supports. This mixture shrinks in volume over time. The shrinkage produces
a tension in the area where paste was applied and in the surrounding areas, causing stress in
portions of the original wood that renders it weaker than before the application of the paste.
6 These crossbars were mounted in such a way that the attachment blocks also reinforced the
join between the upper and lower parts of the support.
7 In the adjustment of the frame junctures, the various pieces had been cut directly on the
painting, during which procedure the serrated blade scraped the painted surface.
8 The use of wedges for rejoining gaps and cracks in panels has already been treated by
Giovanni Secco Suardo in the rst chapter of his manual Il restauratore dei dipinti (1866). This
system is still valid for the restoration operations described here. This method makes it pos-
sible to realign the disconnected edges and rejoin them perfectly through part or all of the
entire thickness, depending on the support condition, without tampering with the preparatory
and paint layers. With regard to destroying original material, it is important to limit such treat-
ment reasonably to the areas of existing degradation: the smaller the area of wood treated, the
greater the stability of the intervention. The wedges positioning, both with the grain and with
the annual rings, is important in obtaining the best stability between the panel and the added
inserts. The wedges must be placed radially with respect to the plane of the support.
9 The correction of the overall curvature was carried out after the V-shaped tracks were cut, so
as to prevent the back edges from pressing against each other during the correction and
thereby spreading apart the painted surface.
10 The mechanism consists of a brass shoe in the form of a closed U at right angles; it is 3 cm
long and 1 cm high and is held to the support with a screw. If the consistency of the wood is
poor, a double-threaded brass bushing is used for attachment to the support; the external
threading anchors to the panel and the internal threading receives the screw that holds the
brass shoe to the support. This 1 cm diameter bushing is 1.5 cm high, according to the charac-
teristics of the support, and is screwed and glued to the planks with epoxy resin. Inside this
brass shoe glides a nylon slide with a screw at the center that is inserted into the thickness of
the framework; the hole has a cylindrical brass sleeve that receives a spiral spring inside it; the
upper end of the bolt is adjusted by a nut.
11 This type of interventioneven if it presents some diculties with the possible removal of
the wax from the back of the panelwas considered useful given the precarious stability of the
painted surface.
12 The vacuum technique is very important for the consolidation of the preparation and paint
layers; its use requires a deep understanding of the application method, which is related to the
solidity of the support and to the particular characteristics of the paint.
13 The method used to carry out this measurement took advantage of a measuring device
suciently sensitive to plot the movement of the object in response to variations in RH. The
gauge is a suitable device for obtaining these measurements. Such an instrumenthaving a
centesimal scale and a useful eld of 10 mmwas modied for use by the attachment to each
of the two ends (xed and sliding) of a perpendicular support ending with a 3.5 mm steel
sphere. On the back of the support, the reference couples were attached in a stable and easily
removable manner. These consisted of nuts with an internal hole ground to 3.5 mm, glued
with epoxy resin to three-prong thumbtacks that allowed the terminals of the sphere to be
housed stably.
339 Tnt Rtsrorari o or Pati Pai ri o Surrorrs . Soxt Cas t Hi srori ts
14 The paintings planks measure, starting from the left of the painted surface: 50.5, 47.5, 45, 50,
47.5, and 36 cm.
15 The oating tenon is a rectangular element of hardwood, often walnut; it works as a connec-
tion and reference point between the planks during gluing of the joins. Floating tenons are
inserted in the housings without being glued and are held in place with one or two dowels per
side inserted into the thickness of the planks. A wooden peg also has the same function,
although it has a circular section.
16 For crossbar removal, the painting was laid at on a wooden grill. With a portable circular
saw, the crossbars were cut longitudinally from below through the entire thickness without
damage to the support.
17 As described in the previous intervention, this method attempts to prevent the creation of
fracture lines by the positioning of small blocks of the same type of wood, united together
and staggered along the length.
18 First, blocks are inserted to four-fths of the depth of the track in the original support. After
the glue has dried, the upper face of the blocks is planed, a procedure that widens the track
about 11.5 cm in the longitudinal direction of the wood bers. Thus, the block applied to
complete the plane with the support will be adhered to the surface of the panel in the direc-
tion of the bers.
19 This method requires the planing of the edges by a slight angling of the utensil toward the
back without its touching the paint edge. With such a system, the reunited planks create a
V-shaped space to receive the wedge-shaped block.
20 The wedge, the central element of this operation, must follow specic criteria: wood selec-
tion, grain orientation, and leveling of the edges of the painted surface. Adjustment of the
wedge in the housing is carried out in the traditional manner.
21 The term traditional method here refers to the opening of V-shaped tracks with a chisel,
correcting the edges, straightening the faces, and adjusting the wedge in the V-shaped track.
Secco Suardo, Giovanni
1866 Il restauratore dei dipinti. Reference
340 Cas t e l l i
T
o txriai rnt arrroacn to the structural conservation of
panel paintings described in this article, the author believes that it
may be more useful to chronicle a single, complex intervention
rather than catalogue the range of technical solutions employed for
specic problems, because he considers the decision-making process
related to a particular intervention to be the most critical factor determin-
ing its success. Obviously, an accomplished level of woodworking skills,
knowledge of the properties and behavior of wood, and a general techni-
cal and mechanical versatility are important, but ultimately they are not
enough to ensure the suitability of a proposed treatment.
The danger of approaching a structural intervention (or any
restoration) from a purely technical point of view is that of unwittingly
causing some kind of aesthetic shift inappropriate to the work of art in
question. Many transfers, for example, can be considered technically suc-
cessful but may have been executed at the expense of certain textural qual-
ities in the surface. Conservators have sometimes been unqualied to
judge the extent to which these subtle shifts compromised the overall aes-
thetic of the object and, ultimately, much of its meaning. Critical aesthetic
judgment should be an essential component of any conservation project,
as it provides the only means to evaluate the appropriateness of a pro-
posed treatment in proper context. This ability is continually developed by
broadening ones general art-historical knowledge, by closely examining
and comparing similar works of art (particularly those in excellent states
of preservation), and by learning how to predict the natural aging behav-
ior of materials under various conditions. Building this kind of knowledge
sharpens ones ability to deduce the fabrication methods and treatment
history of an object accurately, prior to intervention; it also helps in pro-
jecting what kind of improvement can reasonably be expected.
Conservators who believe that aesthetic choices are subjective and
therefore inappropriate relinquish their responsibility to understand the
object in a larger context. Because visual acuity and the complexities of
cultural context are limitless, ones current level of understanding is always
inadequate; consequently, there is a danger inherent in all interventions.
Since any intervention can potentially disrupt the aesthetic and physical
integrity of the object, conservators are bound to consider both of these
aspects in order to minimize the risk of causing some inappropriate shift.
In general, post-treatment environmental conditions should also
be a factor in deciding the extent of a proposed treatment. For example, a
341
George Bisacca
Structural Considerations in the Treatment of a
Nativity by Francesco di Giorgio Martini
cradle that has blocked and caused splits in the panel in the past but is now
housed in a stable environment may require no treatment, provided that
the painted surface is acceptable and the exhibition conditions will not fur-
ther aggravate the state of the panel. Finally, some consideration should
be given as to whether the amount of risk involved in a proposed interven-
tion is justied by the amount of projected gain.
In 1964 Federico Zeri published an article in Bollettino darte linking a
Nativity by Francesco di Giorgio in the Metropolitan Museum of Art in
New York with a fragment, God the Father with Angels, in the National
Gallery in Washington, D.C. (Figs. 1, 2) (Zeri 1964). He recognized that,
given the size of the Metropolitan Nativity (62.2 59.1 cm) and its likely
date, the rectangular format was improbable. He suggested that, stylisti-
cally, the upper portion of the panel required an arched top, and that
iconographic considerations would have dictated the representation of
God the Father giving a blessing or, at the very least, some sort of compo-
sitional closure. The Washington panel furnished precisely these elements.
The oval format of the Washington panel had long been consid-
ered suspect. In fact, various additions to the lower edge could be dis-
cerned with the naked eye beneath the overpainted sky. Close examination
of the Metropolitan panel revealed a horizontal addition of approximately
10 cm along the top edge.
1
Consequently, Zeri hypothesized that the two
Overview
342 Bi s ac c a
Fi gure 1
Francesco di Giorgio Martini, Nativity,
147172. Tempera on panel, 62.2 59.1 cm.
Metropolitan Museum of Art, New York.
panels were actually fragments of the same picture. Zeris reconstruction
was generally accepted in the literature and later conrmed by X radiogra-
phy. It was not, however, until the planning stages of the Metropolitan
exhibition Painting in Renaissance Siena: 14201500, held in 1988 in honor of
John Pope-Hennessys seventy-fth birthday, that a decision was made to
exhibit the pictures together.
The treatment of the panels evolved from the initial idea of minor
cleaning and corrective retouching for the purpose of exhibiting the pic-
tures side by side, to a major cleaning and structural intervention, which
ultimately included the reconguration of the Washington panel into a
lunette, the permanent rejoining of the two panels and, nally, joint own-
ership between the two museums.
The scope and objective of the intervention broadened several
times during the process because of the emergence of new information
that continually expanded the understanding of the work as a whole. New
physical evidence uncovered at various stages pointed toward the need for
increasingly extensive interventions that the conservators and curators
concluded were justied by the prospect of real aesthetic gain with the
least conjecture. At each step, intervention was limited to the minimum
necessary to achieve a clearly attainable goal, based on structural and
aesthetic integrity within the given context. As the context changed, the
permanent rejoining became a more and more logical alternative; and it
stands as a credit to the conservation, curatorial, and administrative stas
343 Trtarxtr or a Nari vi r t Fracts co ni Gi oroi o Marri i
Fi gure 2
Francesco di Giorgio Martini, God the Father
with Angels, 147172. Tempera on panel,
36.5 51.8 cm. National Gallery of Art,
Washington, D.C.
of the National Gallery and the Metropolitan that the needs of the object
were allowed to prevail at every turn.
The rocks and bricks depicted in the addition to the Metropolitan panel
had always appeared aesthetically unsuccessful. In light of the upcoming
exhibition in which the two pictures were to be exhibited side by side, the
Metropolitan decided to reconstruct the work more accurately in relation
to the artists original intention. The reconstruction was based on the
information available in the radiograph taken in Washington, D.C., in the
1960s, shortly after the appearance of Zeris article. Washington produced
a new radiograph on more modern equipment that greatly claried the
original depiction, and the Metropolitan made adjustments that more
accurately reected the new information.
The National Gallery had not previously considered removing
the overpaint in the sky because the outcome was likely to be even more
confusing and dicult to resolve aesthetically. Since the new radiograph
revealed much sharper and more extensive detail than was previously
visible, however, the National Gallery decided that it was now worthwhile
to remove the overpaint and expose as much of the original surface as pos-
sible (Fig. 3). The result was surprising, because the overpaint in the sky
had obscured much of the exquisite detail in the depictions of the thatched
roof and all of the brickwork in the lower fragmentelements that were
Initial Phase
344 Bi s ac c a
Fi gure 3
Washington panel after removal of overpaint
by Sarah Fisher, head of paintings conserva-
tion at the National Gallery. The opacity
of the overpaint had prevented much of
the detail in the thatched roof and all of the
brickwork in the lower fragment from reading
in the radiograph.
not even visible in the new radiograph. The approximate positioning of
the fragments outside the lower left and right of the oval was now clear.
Enough new information was now available to consider returning the
painting to a lunette format. The respective museums agreed not only
to turn the Washington picture permanently into this format but also
to remove the addition from the Metropolitan picture and to abut the
two pictures one above the other in a single frame made specically for
the exhibition.
The removal of the addition on the Metropolitan picture was relatively
straightforward. The panel had been thinned to approximately 1 cm
overall and heavily cradled (Fig. 4), and while many splits and surface dis-
tortions were present (Fig. 5), the panel showed no signs of recent move-
ment. It was, therefore, decided to remove only as much of the cradle as
was necessary to facilitate the removal of the addition. The grain of the
addition was oriented horizontally, while that of the original panel was
vertical. Close examination of the joint revealed an extremely asymmetri-
cal tongue-and-groove joint (Fig. 6a). The upper lip of the groove section
measured less than 1 mm thick, which is an unlikely conguration. It was
speculated that this must have originally been a symmetrical joint, the sec-
tions of which measured 5, 6, and 5 mm, for a total of 16 mm (Fig. 6b),
which is still too thin for an Italian poplar panel of this size. Based on
comparison with other unthinned Sienese panels of this period and date,
it seems more likely that the original thickness was between 2 and 2.5 cm.
Once the panel was cut in two, it may have seemed unnecessarily thick. At
this point, it was probably partially thinned (to 16 mm), and the addition,
with a symmetrical joint, was added. At some later date, after slight warp-
ing, the panel and the addition were further thinned, probably to obtain a
at surface for the application of the cradle, leaving what for all practical
purposes was a half-lap joint.
Removal of the
Metropolitan Addition
345 Trtarxtr or a Nari vi r t Fracts co ni Gi oroi o Marri i
Fi gure 4, bel ow
Reverse of the Metropolitan Nativity. The
joint of the cross-grain addition can be seen
just below the rst crosspiece from the top.
Fi gure 5
Raking light photograph clearly showing
splits and distortions in the Metropolitan
panel and the smooth surface of the addition.
Because of the inherent weakness of any end-grain bond (espe-
cially with hide glue), it was necessary to pare away only the tongue of
the addition that overlapped the original panel and the 1 mm lip (Fig. 6c).
Then, when the addition was rocked gently, the brittle hide glue fractured
neatly along the joint without disturbing the original panel.
Close examination of the Washington panel seemed to indicate that the
abrasion around the edges of the fragments had been caused largely by
an attempt to level uneven surfaces after the fragments had been glued in
place (Fig. 3). Again, the initial idea here was to separate the fragments
and reposition them while causing as small an alteration to the existing
structure as possible.
The cradle, applied by Stephen Pichetto in 1944,
2
was typical of
the method he almost invariably employed (Figs. 7, 8). He thinned the
panel to approximately 5 mm and then laminated it to a mahogany panel
of 1 cm thickness oriented in the same grain direction. He then neatened
the ragged edge of the original poplar panel by the addition of a very
thin mahogany band around the perimeter to match the laminated layer
(Figs. 8, 9). A cradle was then attached; it consisted of mahogany members
oriented in the same grain direction as the panel, with maple crosspieces.
The plan was to dismantle only the section of cradle behind the
fragments and then to attach a platform (extending from the two wide
lobes on the left and right of the cradle) on which to position the frag-
ments and reconstruct the lunette (Fig. 7).
Two crosspieces at the lower edge were removed, and the
mahogany cradle members were sawn through and pared away. The joint
line was then marked precisely on the reverse and cut about halfway
through the mahogany laminate. The saw cut was made perpendicular to
the picture plane, and some additional mahogany was then further pared
away on the fragment side to form a V-shaped opening, which enabled a
better view of the bottom of the cut (Fig. 10). By repetition of this
process, sawing and carving slowly advanced the cut without the risk of
cutting into the original poplar.
Removal of the
Washington Additions
346 Bi s ac c a
Fi gure 6ac
Metropolitan panel prior to intervention.
Note (a) the extremely thin lip of the groove
section toward the back of the panel, and
(b) the presumed thickness of the Metropolitan
panel at the time the addition was added; the
dotted line indicates the thickness at the time
the cradle was applied. After removal of the
cradle (c), the shaded area was carved away by
hand, leaving only one weak, end-grain bond
between the addition and the original panel.
a
b
c
347 Trtarxtr or a Nari vi r t Fracts co ni Gi oroi o Marri i
Fi gure 7
Reverse of the Washington panel prior to
intervention.
Fi gure 8
Side of the Washington panel. The cradle is
approximately 2 cm thick, followed by a
mahogany backing 1 cm thick and nally by
the original panel, approximately 5 mm thick.
Fi gure 9
Edge of the Washington panel after removal
of ll material. The thin mahogany strip can
be seen around the perimeter.
Fi gure 10
Carving into the mahogany laminate of the
Washington panel, in order to read the depth
of the saw cut.
Once the original poplar panel could be seen at the bottom of the
cut, the panel was turned over, and the mahogany edging strip was cut in
correspondence with the saw cut. Since the grain direction of the panel
ran opposite to the fragments, as was the case with the addition to the
Metropolitan panel, the hide glue bond was tenuous. Gentle rocking pres-
sure was enough to fracture the glue easily and separate the fragments
(Fig. 11). The bulk of the mahogany remaining on the fragments was then
removed with a band saw, leaving approximately 1.5 mm attached to the
poplar (Fig. 12). The band saw was used because it exerted far less down-
ward pressure on the paint lm than the amount that would have been
required to carve away the remainder of the mahogany. The operation
took only a few seconds and, with a well-tuned band saw, required nearly
no pressure and little risk.
The next step was to pare away the remaining mahogany from the
fragments. During this step, a layer less than 1 mm thick of alternating
bands of mahogany and poplar was encountered (Fig. 13); it formed a con-
tinuous layer between the original panel and the mahogany laminate in
the entire Washington picture. (It was oriented in the grain direction
and consequently ran cross-grain only under the small fragments.) The
purpose of this layer is not understood at present, but it may have had
something to do with adhesive compatibility. The poplar would perhaps
adhere better or react similarly to the poplar panel, and the converse
would be true for the mahogany. These alternating bands were literally
paper-thin and could easily have gone undetected. The existence of these
bands is interesting, given the fact that, in contrast to most cradles, those
applied by Stephen Pichetto usually function well, even after fty years.
This small detail may contribute in some way to their success.
348 Bi s ac c a
Fi gure 11, above
Separation of the fragments of the
Washington panel.
Fi gure 12, above ri ght
Fragments after removal of the bulk of the
mahogany from the Washington panel with
the band saw. The 1.5 mm thickness of
mahogany can be seen attached to the blue
fragment in the foreground.
Fi gure 13, ri ght
Alternating bands of mahogany and poplar
between the original panel and the mahogany
backing of the Washington panel.
During the removal of this layer, another detail came to light that
substantially altered the plan for completing the lunette. After removal of
the alternating bands on the small fragments, it was found that the origi-
nal poplar extended beyond the painted areas at the pointed ends of the
fragments (Figs. 3, 9). It had been previously assumed that these areas had
been lled with additional scraps of old poplar, but, instead, the poplar
was continuous.
Since the curve traced by the edge of the painted surfaces of
the fragments contained the lip or barb characteristic of the perimeter
of painted panels with engaged frames, it was deduced that the tips of
the fragments had not been painted originally and were instead part of the
panel onto which the original framing elements had been axed before
gessoing, as was common in the late quattrocento (Fig. 14a, b). A prelimi-
nary arrangement of all four fragments based on the convergence of lines
in various design elements showed that the unpainted tips of the small
fragments extended beyond the painted surface of the main fragment
(Fig. 15). Obviously, they could not be trimmed o because they were the
only evidence clarifying the original structure of the panel; instead, wood
was added around the perimeter to encompass these tips. This addition
would not be visible ultimately, because it would eventually be cropped by
the frame, but it was considered necessary, nonetheless, to protect the frag-
ments better while more clearly indicating their function. To attach the
addition, however, further alterations would have to be made to the cradle.
Since the bottom of the cradle was already altered, and the author
was also planning to alter the lobes at the left and right, altering the upper
curve as well would mean that the only undisturbed section of cradle
would be a small area in the center. In light of this new information,
retaining the cradle became a less logical alternative. After consultation
with the conservation department of the National Gallery, it was decided
to remove the entire cradle but leave the mahogany laminate attached to
the original poplar panel. Although this alternative appeared unnecessary
349 Trtarxtr or a Nari vi r t Fracts co ni Gi oroi o Marri i
a b
Fi gure 14a, b
Francesco di Giorgio Martini, Madonna and
Child with Angel, 1474. Tempera on panel,
74.5 49 cm. Pinacoteca, Siena (cat. 288). The
panel and frame are integral to this work (a).
Gesso was applied continuously over the
frame and panel before gilding and painting.
The exploded view (b) of the engaged framing
elements shows that if they were removed, the
uncoated portion of the panel would be visi-
ble. The unpainted tips of the small fragments
in the Washington panel indicate that this was
the original method of construction.
at the beginning of the intervention, it became more obviously logical and
ecient after the discovery of the unpainted tips. All additions would be
built onto the mahogany without disturbance to the poplar panel. After
the cradle was removed and the back scraped clean, a track was routed to
half the thickness of the mahogany, and new mahogany pieces were tted
to extend that plane to accommodate the fragments, including their pro-
truding tips (Figs. 16, 17). After the exact placement of the fragments was
decided, the areas that were completely missing would need to be built up
from the mahogany to the level of the gesso preparation (see Figs. 8, 9).
Very old poplar brought from Italy was used for this purpose in order to
maintain a consistent structure (Fig. 18). After the poplar collar was glued
to the mahogany with a polyvinyl acetate (PVA) emulsion (Fig. 19), the
fragments were set into the cutouts in the collar and precisely aligned with
the surface of the main part of the panel, both along and across the grain.
Rabbit-skin glue thickened with calcium carbonate was used as an adhe-
sive to ll any gaps caused by adjusting for surface level. Other adhesives,
such as Ciba-Geigy Araldite 1253 carvable paste, have excellent gap-lling
properties as well as much longer curing times; however, the traditional
350 Bi s ac c a
Fi gure 16, bel ow
Fitting of the mahogany extensions for the
Washington panel.
Fi gure 17, bel ow ri ght
Washington panel after the completion of the
mahogany additions.
Fi gure 15
Preliminary arrangement of all fragments.
organic adhesive was selected because it is more easily reversible.
3
The dis-
advantage of its quick setting time was minimized by repeating the clamp-
ing procedure dry several times until the same results could be achieved
consistently, accurately, and quickly. The perimeter was then drawn and
trimmed on the band saw (Fig. 20).
Two crosspieces of the Florentine type described elsewhere (see
Rothe and Marussich, Florentine Structural Stabilization Techniques,
herein) were then fabricated and applied (Fig. 21). The contact faces of the
crosspieces themselves, as well as the small retaining pegs that hold the
crosspieces, are machined to an angle of 22.5. The two rows of pegs cre-
ate a sort of dovetail track within which the crosspiece can slide, allowing
for any lateral expansion and contraction of the panel. The trapezoidal
shape of the crosspiece also permits convex exing of the panel, and the
small contact faces of the pegs minimize friction against the crosspiece,
making it virtually impossible for it to bind. The pegs were attached to the
panel with Ciba-Geigy Araldite 1253 carvable paste, the gap-lling proper-
ties of which make it possible to set the contact between the peg face and
crosspiece precisely, while the resin adequately compensates for any irregu-
larity between the peg bottom and the panel. If inordinate pressure were
eventually to accumulate in the panel from warpage, the small pegs would
tend to delaminate rather than cause the panel to split.
This type of secondary support (later abandoned because of
developments described on the following pages) was applied only to a
panel that had previously been thinned considerably for warp reversal or
for the application of a cradle. It would generally not be used for a panel
351 Trtarxtr or a Nari vi r t Fracts co ni Gi oroi o Marri i
Fi gure 18, above
Poplar collar used to build up the missing
areas to the surface level of the original
poplar panel of the Washington panel.
Fi gure 19, above ri ght
Gluing of the collar to the mahogany backing
for the Washington panel.
Fi gure 20, bel ow
Band saw being used to trim the perimeter of
the Washington panel.
Fi gure 21, bel ow ri ght
Reverse of the Washington panel after appli-
cation of the crosspieces.
that had retained its original surface, because the support was considered
too great an aesthetic intrusion.
The missing areas were then gessoed and inpainted by Sarah
Fisher, head of paintings conservation at the National Gallery (Figs. 22, 23).
Both pictures were then butted together without glue in a single frame
made for the exhibition (Figs. 24, 25).
This arrangement proved successful enough to prompt the two
institutions to agree to the permanent rejoining of the panels subsequent
to the exhibition and joint ownership thereafter.
Among the problems presented by the permanent rejoining, the most
dicult to resolve was the discrepancy between the beautiful, uniform
surface of the Washington panel and that of the Metropolitan panel,
which displayed such problems as several open splits, warps, and planar
distortions, most of which were related to the cradle (see Fig. 5). Although
the condition of the Metropolitan panel was less than ideal, it was none-
theless stable, given the satisfactory environmental conditions within the
museum. While the aesthetic improvement of the surface had always been
an attractive idea, it was felt that the subtle aesthetic gain did not justify
the extensive structural treatment to which the panel would have to be
subjected. Now, however, in light of the permanent rejoining, the relation-
ship between the upper and lower surfaces seemed important enough to
justify the intervention.
Permanent Rejoining
352 Bi s ac c a
Fi gure 24, ri ght
The two pictures abutted together without
glue in a single frame made for the 1988 exhi-
bition Painting in Renaissance Siena: 14201500.
Fi gure 25, f ar ri ght
Reverse of the panels during the exhibition.
The Washington panel (above) is wider than
the Metropolitan panel because of additions
made to the perimeter.
Fi gure 22, above
Washington panel after application of gesso.
Fi gure 23, above ri ght
Washington panel after inpainting by
Sarah Fisher.
The cradle (Fig. 4) not only limited access to the splits but also
impeded the improvement of the surface alignment and adjustment of the
overall curvature of the panel. These problems were compounded by a
thick layer of wax that had been poured hot over the entire cradle and
panel, probably in the 1950s.
The wax and cradle were removed. The splits were repaired using
the Florentine wedge method, which consists of the following procedure:
First, V-shaped tracks are cut as narrowly as possible along the splits,
and wedge-shaped pieces of wood are then tted with extreme precision
and glued into the tracks. The wood used is of the same type as the panel
and as close in age as possible. An attempt is made to match the grain
direction, cut, and even the degree of worm tunneling, so that the repair
does not exert a greater or lesser structural force within the panel. (For
examples of this technique, see Rothe and Marussich, Florentine
Structural Stabilization Techniques, herein.)
This controversial method was developed for a number of rea-
sons. Of course, whenever possible, simple splits that t together well
should merely be reglued; however, in many cases they are too tight for
glue to be introduced to the full depth. As a result, they continue to move
near the paint surface, causing new lls to reopen and splits to continue to
lengthen. In the case of older splits, some are considerably more open on
one end than on the other and cannot be closed without excessive pres-
sure; in such cases, lling them with relatively large amounts of adhesive
would become necessary. Others, because they have warped dierently on
both sides of the split, have complicated surface leveling problems and
other planar distortions. And others, because of repeated treatments in the
past, are lled with wax, dirt, varnish, gesso, and organic and inorganic
adhesive residues that impede accurate regluing.
By cutting a narrow, V-shaped track, one gains full access to the
entire depth of the split while removing any extraneous material and
exposing pristine gluing surfaces for a better adhesive bond. Surface curva-
ture and level can be precisely adjusted in very short segmentseven one
wedge at a timeensuring highly controlled results. The precision of the
t can reduce the amount of adhesive necessary by several hundred per-
cent. By the tting of short wedges, the faces of each segment of the
V-shaped track can be readily prepared perfectly at, and irregular splits
can be followed with greater accuracy. Moreover, if the wood of the
wedge were to have any tendency to move dierently from the panel, its
strength would be minimized by the interruption of the cell chains due to
the short lengths of the wedges: it is unlikely that individual wedges could
do more than simply follow the movements of the panel.
The controversial aspect of this method is, of course, the removal
of original material. Two factors come into play in this regard. One is the
undeniable primacy of the painted surface and its ability to function or
convey its particular pictorial meaning. The second is the contribution the
panel makes toward the overall aesthetic of the work of art as an object,
including the practical information that can be gleaned from tool marks,
dowel holes, edges, metal attachments, and so forth; this evidence can
shed light, for example, on fabrication techniques, placement within an
altarpiece, and original collocation, and it must be scrupulously respected.
These two factors must be considered together in the planning of
the extent of any structural intervention. The situation is substantially
dierent, however, when a panel has been thinned and cradled. Any
353 Trtarxtr or a Nari vi r t Fracts co ni Gi oroi o Marri i
aesthetic aspect or technical information contributed by the original wood
surface has already been eliminated. Therefore, the decision to remove a
small amount of material that had never been visible from the exterior, in
a process that could greatly facilitate the aesthetic improvement of the
painted surface as well as the future stability of the panel, can be justied.
If, to take a hypothetical example, a panel were to lose both its back sur-
face as well as its painted surface, would the core material retain any value
as a work of art or even as a historical document?
It should be stressed that the removal of original materialeven
that which was never visible on the surfaceremains a radical decision
and should not be undertaken as a matter of course. The tting of these
wedges is a dangerous operation and, unless it is very precisely executed,
oers little advantage over simpler methods of regluing. When executed
precisely, however, it produces a repair of exceptional stability and durabil-
ity while allowing uniform, uninterrupted expansion and contraction
across the panel. It also permits extremely accurate surface-level and cur-
vature adjustments with minimal aesthetic compromise.
An interesting solution for Botticellis Man with a Medallion in the
Uzi Gallery was recently presented by Ezio Buzzegoli and Marco Marchi
of the Soprintendenza per i Beni Artistici e Storici di Firenze e Pistoia
(Buzzegoli, Marchi, and Scudieri 1993). In this example, the panel retains
its original surface, but a split traveling upward from the bottom was caus-
ing the pastiglia medallion held by the sitter in the picture to fracture.
With a scalpel, conservators removed the skin of wood around the split
in one continuous piece, tted the split with wedges, and reattached the
skin so as not to disturb the overall aesthetic.
Attachment of the two panels
After all the splits in the Metropolitan panel were tted with wedges,
thereby improving the surface leveling and overall curvature, the two pan-
els were then aligned (Fig. 26). The top of the Metropolitan panel already
had a routed track from the attachment of the old addition. It was decided
to rout a similar track into the mahogany of the Washington panel with-
out cutting into the original poplar (Fig. 27). Short poplar blocks were
then made to bridge the two panels. The blocks were tted and glued a
few at a time, beginning at the center, so the paint surface could be care-
fully leveled as each piece was glued in place (Fig. 28). Before each piece
was glued, the end-grain joint between the panels was lled from the back
with gesso to produce a tighter t. Each xed piece provided a point of
leverage from which to level the next pieceand so on, until the track was
completed. Rabbit-skin glue thickened with calcium carbonate was used
again as an adhesive.
The sides of the Washington panel were wider than those of the
Metropolitan panel because of the additions that were made around its
perimeter (Figs. 25, 29). No additions had been made to the Metropolitan
panel because, although the same amount of wood was missing around
its perimeter, as long as the panel existed as a separate entity, there was
no pressing need to reconstruct it. Besides, it would eventually only be
cropped by the frame. Other evidence (the raised lip or barb at the edges
of the painted surface) made it clear what was missing, and this was
considered sucient. The decision to add the missing wood in the
354 Bi s ac c a
Fi gure 26
The Washington panel (left) and the
Metropolitan panel, aligned in position for
permanent rejoining.
Fi gure 27
Planing of the track between the two panels.
Note the gesso lling added into the slightly
irregular joint at the center.
Washington panel was motivated primarily by the need to nd a solution
that would physically protect the tips of the fragments without falsifying
the aesthetic of the object. Now that the panels were permanently
rejoined, it made sense to add the missing strips to the sides of the
Metropolitan panel as well, in order to simplify the perimeter, reect the
original fabrication method, and strengthen the entire construction.
It was decided not to continue the addition across the bottom
edge of the picture because it was considered unnecessary. Not only would
the end-grain attachment present its own problems, it would not in itself
resolve any other problem.
The back surface of the Metropolitan portion of the panel had
been scraped in order to remove all wax and glue residues, and it was now
judged to be potentially highly reactive to humidity uctuations. A coat
of Acryloid B72 was applied; this product is not totally impermeable but
merely slows down the moisture exchange rates.
Two crosspieces similar to those already applied to the Washington
portion were fabricated and attached (Fig. 30). The coat of Acryloid B72
applied as a moisture barrier would also facilitate the release of the retain-
ing pegs of the crosspiece system in the event that too much stress were to
accumulate at some point in the future.
The two institutions formally agreed to alternate custody of the
newly rejoined panel every ve years (beginning in Washington, since it
had rst been exhibited briey at the Metropolitan for the exhibition of
Sienese Renaissance painting). The interval was considered reasonable
given the proximity of the two museums. A custom-designed crate
equipped with cushioning for shock as well as vibration absorption and
thermal insulation was then provided. Whenever the painting travels, it
will be accompanied by museum couriers and transported via truck with
air-ride suspension and climate control, for better control of cargo han-
dling and climatic variables than if the painting were transported by air.
One nal alteration was made to the secondary support upon its
return to New York in October 1994. Although the four Florentine-type
355 Trtarxtr or a Nari vi r t Fracts co ni Gi oroi o Marri i
Fi gure 28
Gluing of the blocks.
Fi gure 29, ri ght
Upper right portion of the joined panels.
Fi gure 30, f ar ri ght
Reverse of the joined panels after application
of the crosspieces on the Metropolitan panel.
Note the strips newly added to the sides of
the panel, as well as the continuous narrow
strip of poplar that forms a bridge to the
Washington portion, greatly increasing the
stability of the joint.
crosspieces described above functioned adequately, they were substituted
with the type of secondary support rst published by Ciro Castelli and
Marco Ciatti (1989). This system consists of a strainer that follows the
perimeter of the panel exactly and, in this case, has two xed crosspieces
and one xed vertical member. The strainer is simply held in place by
springs, xed to the strainer on one end and attached on the other end to
small blocks of wood oriented in the grain direction; these blocks are, in
turn, spot-glued to the back of the panel (Figs. 31, 32). The spring is not
xed rigidly to the small block but instead slides freely within a predrilled
hole, allowing for expansion and contraction of the panel, as well as con-
vex exing or even straightening. The bottom edge of the strainer has a
small lip that protrudes to accommodate the thickness of the panel and
prevents the weight of the panel from fatiguing the springs over time.
This system oers several advantages over the traditional
Florentine-type crosspieces previously employed, especially in the case of
very thin panels. The strainer protects the fragile perimeter and oers
greater resistance to torquing and better overall stability, while more
closely approximating the original thickness of the panel and making it
easier and safer to handle. The system also reduces the surface area
adhered to the panel while distributing the support more regularly and
without adding any weight to be supported by the panel. It also allows
more localized, independent movement of any specic area of the panels,
and the spring tension can be calibrated to take into account the species of
wood, thickness, cut, degree of worm infestation, and past treatments.
4
Finally, if the screws that attach the springs to the strainer are
recessed, a lid or cover can be tted over the back. Not only does this
cover oer protection, it also creates a microenvironment that can buer
humidity uctuations. Furthermore, silica gel tiles can be attached to
the inside of the lid between the various crosspieces. It should be noted,
however, that this solution makes no attempt to function as a climate-
controlled vitrine. There is no glass in front of the picture, and the sides
are not sealed. Prolonged exposure to low humidity will produce the same
eects as the absence of silica gel. However, the semicontrolled environ-
ment can constitute a substantial buer to humidity uctuations, even
eliminating movements of the panel that would be caused by daily humid-
356 Bi s ac c a
Fi gure 31, ri ght
Reverse of the panel with the spring tension
strainer in place.
Fi gure 32, f ar ri ght
Spring mechanism shown in place.
ity oscillations in the range of 1015%. Essentially, the panel reacts more
or less as though it were unthinned.
Silica gel tiles were not added to the Francesco di Giorgio Martini
work; however, had the panel been more quickly reactive or had the envi-
ronmental conditions within the two institutions been less stable, they
would have been a likely option (and they remain an option in the future).
Whenever this type of secondary support is used, care should be
taken to secure the object in its frame by means of some kind of exible
clip. If it were xed rigidly and the panel were to increase its convex warp,
the necessary movement would otherwise be blocked by the frame rabbet.
Although the various phases of treatment of these panels oer no
technical innovations, the project as a whole demonstrates the degree to
which overall context plays a determining role in assessing the appropri-
ateness of any proposed treatment. In this instance, solutions were repeat-
edly modied throughout the treatment process to accommodate new
physical and contextual information that came to light during the course
of the intervention (Fig. 33).
357 Trtarxtr or a Nari vi r t Fracts co ni Gi oroi o Marri i
Fi gure 33
Metropolitan and Washington panels after
permanent rejoining.
The author would like to thank Keith Christiansen, Jane Wrightsman
Curator for Italian Paintings at the Metropolitan Museum of Art, New
York, for his inspired and enthusiastic collaboration; Tom Wilmering, head
of furniture conservation in the Objects Conservation Department of the
Metropolitan Museum, and Ciro Castelli of the Opicio delle Pietre Dure,
Florence, for their valuable technical advice; and Sarah Fisher, head of
paintings conservation, National Gallery, Washington, D.C., and Laurent
Sozzani, then an intern at the Metropolitan Museum, for their dedicated
work on the painting surfaces.
1 A photograph taken in 1897 in the les of the Metropolitan Museum already records the addi-
tion to the top of the Metropolitan panel.
2 Stephen Pichetto was trustee and/or curator of the Kress Collection from 1932 until his death
in 1949. He maintained a large conservation studio with several employees, and many impor-
tant pictures purchased in America during this period were treated in his studio. A man named
Angelo Fatta was apparently responsible for the thinning and cradling of panels under
Pichettos direction.
3 In conservation practice, reversible is often synonymous with soluble. Obviously, the Ciba-Geigy
Araldite is not reversible in this sense. In many cases the solubility of an adhesive would not
be physically possible or even desirable. For instance, attempting to dissolve a water-soluble
adhesive sandwiched between wooden elements beneath a gesso ground, all of which are
hygroscopic, would have disastrous results. Often, as in the present example, mechanical rever-
sal of an insoluble adhesive would be preferable to any attempt at dissolving an adhesive layer.
Since rabbit-skin glue becomes brittle, it is possible to carve down to the glue line and scrape
away the glue without further damage to the original panel.
4 These variables would be very dicult, if not impossible, to quantify. Consequently, spring
tension must be set according to an empirical understanding based on knowledge accumulated
from the handling and exing of similar panels.
Acryloid B72, Rohm and Haas Co., Independence Mall Street, Philadelphia, PA 19105.
Araldite 1253 carvable paste, Ciba-Geigy Corporation, 4917 Dawn Avenue, East
Lansing, MI 48823.
Buzzegoli, Ezio, Marco Marchi, and Magnolia Scudieri
1993 Luomo con la Medaglia del Botticelli e il Morgante Nano del Bronzino. OPD Restauro
5:2327, 6768.
Castelli, Ciro, and Marco Ciatti
1989 Proposta di intervento su particolari supporti lignei. OPD Restauro 1:10811.
Zeri, Federico
1964 Un intervento su Francesco di Giorgio Martini. Bollettino darte (1):4144.
References
Materials and Suppliers
Notes
Acknowledgments
358 Bi s ac c a
T
ni s rtii tr, dated to around 1515 and part of a retable of
unknown provenance in Ulm, has been attributed to Martin
Schaner (1477/781546/49), an artist active in Ulm and the duchy
of Swabia (Fig. 1). Its composition, reminiscent of an engraving by Martin
Schongauer (ca. 143091), could be a model for other Swabian reliefs of
the Annunciation of the same period (Sprinz 1925).
The relief, on limewood, is 104.8 cm high and 118.1 cm wide.
Mary is represented in front of a tent-shaped baldachin kneeling at her
prayer stand; she is holding her cloak with her right hand, and her left is
resting on her open prayer book. Marys eyes are cast down pensively.
Gabriel has appeared on her left; he is holding a scepter in his left hand,
which also lifts the curtain to open the baldachin. In the center of the
background is a vase with lilies (Eckhardt 1982).
359
Frdric J. M. Lebas
The Cradling of a Relief of the Annunciation
Attributed to Martin Schaner
Fi gure 1
Martin Schaner (attrib.), Annunciation,
ca. 1515, before restoration. Relief on panel,
104.8 118.1 cm. Ulm, Germany.
The relief is composed of four vertical slab-cut boards, reversed
and glued together (Marette 1961). It was reinforced with seven 3 cm thick
limewood boards (ve uprights and two crosspieces), apparently glued and
held together by forty screws (Fig. 2). For hanging, two attachments, each
axed by three screws, were added to the upper crosspiece; the frame was
held in place by four long nails inserted into the sides of the cradle. Two
3 mm wide cracks had run up the whole length of the panel, while two
shorter ones, each about 50 cm long, started upward from the bottom of
the panel. The wood has been heavily eaten by Anobium punctatum worms.
A prior restoration is indicated by numerous llings of holes, reconstruc-
tions in places with wood ller, and portions of various sizes reworked
with limewood. Tunnels left by xylophagous larvae, visible on the surface
of the relief, suggest that the panel may once have been painted; the small
remaining amount of ground does not allow us to be more denitive. The
maximum thickness of the relief is 10 cm. The cradle, with its two cross-
pieces, created the stresses in the panel that caused the cracking discussed
above. Therefore, the cradle had to be removed.
The relief was placed facedown over a thick pad of foam; the cavities in
the wood were then lled with pieces of the same foam cut to size. The
boards were removed one at a time; they were rst cut into pieces of vari-
ous sizes according to the thickness of the relief and the location of the
cracks. Each piece was then thinned down with a chisel, and the surface
was carefully nished with a scalpel and damp pad, which removed all
traces of animal glue. This work uncovered nails and wooden pegs that
had been inserted from the front of the panel to maintain the reconstruc-
tions. The back of the relief being very uneven and the cradle boards quite
at, the cradle boards had not adhered in all places, and in some areas the
glue was 23 mm thick between the relief and the cradleit is easy to
imagine the stresses these irregularities caused on the surface of the wood.
After cleaning, saw marks became visible on the back, indicating that the
Treatment
360 Le bas
Fi gure 2
Martin Schaner (attrib.), Annunciation. Back
of the relief before restoration.
panel had been thinned by sawing (Fig. 3). The relief, now in several
pieces, had to be glued together again.
As the relief was not solid enough to support itself in its frame, a
new structure had to be built. There were many options. The relief is very
irregular: heavy and thick, especially at the left and right margins, and thin
in the center for almost the whole height. A light support was required,
capable of adapting to the potential movements of the original, including
swelling, shrinking, and convex and concave warping. Moreover, the back
of the panel is very uneven. After a few weeks, during which the relief
was left at without constraints, a cradle design was selected: it was to be
made of small balsa-wood pieces, 10 cm long, 4 cm wide, and 1.5 cm thick,
glued in two staggered layers, with the grain direction following that of the
relief. The size chosen for the pieces was related to the width, height, and
thickness of the relief, as well as to the irregularities of the surface.
The back of the panel was very uneven and had many holes, which
needed to be lled in to even out the surface to some extent. Sheets of
limewood veneer, with the edges thinned down and the angles and edges
rounded, were adhered to the panel with Keimx and clamped. After sev-
eral attempts to ll other holes with various glues, these other cavities were
lled with sifted limewood sawdust mixed with ethyl cellulose glue in a
toluene solution: this produced a ne, soft, and easily worked elastic paste.
Next, the back of the panel was coated with a solution of 10% Paraloid
B72 in toluene, to isolate the panel from the wax used to attach the cradle,
thereby preventing penetration of wax into the panels wood. This wax is a
5050 mixture of beeswax and Lascaux 443-95 adhesive wax (pure beeswax
would not have been strong enough; the adhesive wax would have been too
strong). The wax mixture was heated in a double boiler and brushed on the
back of the panel; the mixture was then warmed with an industrial-type
heat gun to spread it evenly in a thin layer.
The cradle was started in a vertical line in the center of the panel.
The balsa pieces were dipped in the hot wax and arranged side by side as
one might build a wall. However, before they were actually glued, they
were set into place to see how well they t. If there was a gap between
361 Tnt Cranii o or a Rtii tr or rnt Auci ari o Arrri turtn ro Marri Scnarrtr
Fi gure 3
Martin Schaner (attrib.), Annunciation. Saw
marks on the back of the relief.
the panel and the balsa, an extra piece of balsa was shaped to ll in the gap;
if, on the other hand, there was a protrusion on the panel, the block was
shaped or grooved to accommodate the protrusion, allowing the block to
t closely against the relief panel. The blocks were then glued down. Once
the rst layer was nished, it was leveled by planing. The second layer of
balsa was placed in the same direction as the rst but was staggered so that
the joints were not superimposed (Fig. 4). That layer was also planed down
after it was in place. The edges were smoothed all around and the cradle
brushed with a solution of 10% Paraloid B72 in toluene, a coating intended
to ensure a good nish. The rst phase of the treatment was over.
The problem of maintaining the relief in its frame, however,
remained. As the frame was made simply of four lateral gilt-edged boards,
the question arose of how to attach it to the relief. Four boards were
added to the inside of the frame, so that an opening was left in the back.
They were glued and pegged; then cleats were glued on the cradle 1 cm
from the inside of the frame with pure Lascaux 443-95 wax, and springs
were screwed onto the cleats to hold the relief in the frame (Figs. 4, 5).
362 Le bas
Fi gure 4
Martin Schaner (attrib.), Annunciation. Model
showing the construction of the cradle and
the new attachment of the frame.
Fi gure 5
Martin Schaner (attrib.), Annunciation. Back
of the relief after restoration.
The advantage of this structure is that it allows the relief to move freely in
all directions within the frame. The attachments for hanging were axed
to the back of the frame and not, as previously, on the cradling. Finally,
the surface of the relief was cleaned, and the old repairs, putty llings, and
gaps were reintegrated with watercolor.
The restoration was completed on 5 May 1982 and has been
inspected regularly since then. It remains in good condition.
Ethyl cellulose glue (N 50), Hercules Inc., 910 Market Street, Wilmington, DE 19899.
Keimx, Keim Leim AG. and Co., Mechternstr. 57, 5000 Cologne 30, Germany.
Lascaux 443-95 adhesive wax, A. K. Diethelm AG, Ch 8306 Brttisellen, Germany.
Paraloid B72 (in U.S., Acryloid B72), Rohn and Haas Co., Independence Mall West,
Philadelphia, PA 19105.
Eckhardt, Wolfgang
1982 Erwerbungen fr die europischen Sammlungen in den Jahren 198081. Jahrbuch des
Museums fr Kunst und Gewerbe Hamburg 1:14446.
Marette, J.
1961 Connaissance des primitifs par ltude du bois. Paris: Picard.
Sprinz, Heiner
1925 Die Bildwerke der Frstlichen Hohenzollernschen Sammlung Sigmaringen. OJS 77:23.
References
Materials and Suppliers
363 Tnt Cranii o or a Rtii tr or rnt Auci ari o Arrri turtn ro Marri Scnarrtr
B
acxi o rati rai ri os with balsa-wood blocks glued with wax-
resin is an uncommon technique rarely used by the Institut Royal
du Patrimoine Artistique, Brussels (IRPA). In ten years, it has been
applied to only about ten paintings. These were oil paintings on oak panels
from the sixteenth or seventeenth century that needed to be reinforced,
maintained, or constrained for stability or display. At IRPA, reections
regarding this type of treatment have been based on the published descrip-
tions of several authors (Buck 1970; Spurlock 1978; Beardsley 1978; vom
Imho 1978).
Since at IRPA reversibility is considered to be an absolute require-
ment for an adhesive, we opted for damar wax-resin rather than a tridi-
mensional resin. Subsequently, collaborative work with our French
colleagues
1
as well as comparative studies made by students (Habaru
199091; Mori 199293) have enabled us to rene this technique. Each
intervention has led to discussion and research aimed at improving the
technique and adapting it to the specic problem of each panel.
First, the two types of deterioration of painted panels that we
believe justify a balsa backing will be described. The choice of materials
and work method will then be explained.
Frequently, panels are found whose original construction has been altered
by earlier, well-intentioned restorers. The history of treatments for paint-
ing supports has seen many changes in fashion. Some problematic panels
underwent the addition of crosspieces, cradling, or even transfer. To
perform these so-called restorations, the supports were thinned down
or eliminated.
Today these restorations are faulted, rst, for radically transforming
the structure of the work and, second, for proving ineective. Moreover,
the irremediable loss of technological and historical evidence is very unfor-
tunate. These restorations must now often be reversed to save the lifted and
distorted paint layer. Once the additions have been eliminated, we are left
with a work whose support is so thinned down that it is no longer able to
stand by itself.
The backing of these thinned-down panels with balsa-wood
blocks adhered with wax-resin presents a signicant advantagenamely,
the method is easily thermoreversible. Because it is made up of a multi-
tude of waterproof cells whose pure cellulose walls are dicult to perme-
Reinforcement of
Thinned Panels
364
Jean-Albert Glatigny
Backings of Painted Panels
Reinforcement and Constraint
ate, balsa wood is an inert material that is not subject to distortion over
time. Once the panel is backed, it is resistant but not much heavier. The
adhesive used, a mixture (by weight) of seven parts beeswax and two parts
damar resin, is relatively exible, and its adhesive strength is moderate.
These particular qualities, while they contribute security to the panel, are
at the same time the techniques weak point. Paintings treated this way
will require special precautions, especially with regard to mechanical
shocks and high temperatures.
Panel paintings, which dont have a paint layer on their reverse, are often
more or less convex. If the distortions are distributed evenly over the
whole of a panel, the viewer will not be troubled. But the presence of a
single, limited distortion can be so disturbing as to alter the look of the
painting completely. These distortions have internal causesfor example,
the woods nature, density, or method of conversion (the way it has
been sawn). They emerge as a result of poor conservation conditions,
such as serious uctuations of relative humidity (RH) and the constraints
wrought by framing.
Balsa-wood backing glued with wax-resin has been used to main-
tain distorted panels after attening. The method consists of increasing
the water content of the whole panel in an air-conditioned chamber and
locally applying damp compresses over extremely distorted areas. The aim
is to reach a point of balance at which the boards recover their inherent
atness. Once this condition is achieved, the backing is applied to the
whole back of the panel. Balsa-wood backing with wax-resin acts as a
mechanically uniform maintenance device; moreover, it slows down
humidity exchanges. It forces the panel to remain at during the drying
period, a process that completes the treatment and subjects the wood cells
to plastic distortion. The return to an RH of about 60% takes place gradu-
ally, in a matter of two to three months.
Balsa-wood backing, in this case, fullls a provisional function. It
can be removed as soon as a panel is stabilized in an environment where
the RH is controlled. To date, however, as a precaution, such backings
have been left on. Balsa and wax-resin act as barriers against humidity.
The advantage of this technique lies in the fact that if the con-
straint on the drying panel is higher than the adhesive strength of the
wax-resin, the backing will come unglued. In such a case the panel will
reassume some curvature and will not be threatened with splitting.
Among the treated paintings, three thin (less than 2 cm) oak pan-
els that were formerly distorted are currently backed. This way, while
keeping them at a stable RH, we have succeeded in keeping them at. If
the RH were not controlled, the backing would retard the emergence of
distortions, but it would not prevent them. The treatment can be applied
again if necessary.
Balsa wood is commonly used in restoration. It is valued for its extremely
low shrinkage, its light weight, and its waterproof qualities. The kind used
at IRPA comes from Ecuador, and its weight varies from 80 kg m
3
to 290
kg m
3
. The elements used were all of the same density, 170 kg m
3
.
After experimenting with rectangular, square, and hexagonal
blocks, in radial and transverse conversion, we opted for 8 cm squares that
Materials
Constraint and
Straightening of
Distorted Panels
365 Bacxi os or Pai rtn Patis . Rti rorctxtr an Cosrrai r
were 1 cm thick. The sawing was done in a transverse direction on end-
grain wood in order to obtain elements that were as rigid and as easy to
work with as possible.
The adhesive is a mixture of seven parts beeswax and two parts
damar resin. This adhesive is one that has been used for fty years by IRPA
for certain relinings of painted canvases and for consolidations of paint
layers as well. To make the adhesive, raw beeswax is obtained from a bee-
keeper. It is then washed in boiling water and ltered. This weak adhesive,
solid at room temperature and liquid at 60 C, is stable and exible, and
it can be easily dissolved or reactivated. It is also a good barrier against
humidity. It impregnates only the surface of the wood. The reverse of a
panel is traditionally sized with rabbit-skin glue at the time of manufac-
ture. This method of insulation, which prevents penetration of the wax, is
an option to consider before backing.
Before a panel painting is backed, the adhesion of the paint layer is exam-
ined. A facing is applied to the painted surface, the joints and splits of the
support are glued, and the lacunae in the wood are lled.
The painting and the balsa blocks are then brought to the same
RH level. With a brush and spatula, a layer of warm wax-resin is applied
over the entire reverse of the panel, in order to level the irregularities in
the wood surface. The wax-resin mixture shrinks as it cools. To control
the extent of shrinkage, it is applied in thin, successive coats. The balsa
blocks are immersed for a few seconds in the melted adhesive, positioned
on the cooled layer of wax-resin on the panels reverse, and held in place
until the wax cools.
The joints between the blocks are aligned diagonally with regard
to the grain of the boards that form the panels. Two levels of blocks are
glued in this way, the second level being staggered so that the joints are
not superimposed. Experience has shown that the joints are the weak
point in the handling of panels, and therefore, that a rigid support is desir-
able. The most rigid support is achieved with balsa blocks sawn in a trans-
verse direction, then placed in two staggered layers, diagonally with regard
to the panels grain.
A sheet of very thin, long-ber paper (12 g m
2
), such as bamboo-
ber paper, is next glued with wax-resin to the backing. After gluing, this
paper is transparent. It allows for the control of the possible opening of
the joints, and it holds the blocks in place in case of signicant ungluing.
The treated panel must be replaced in its frame, which fullls a
dual function: it distributes strains during handling, and it supports the
painting when it is hung.
A balsa backing was carried out in the conservation-restoration workshop
at La Cambre school, Brussels, where the author teaches. Paul Duquenois,
a fth-year student of painting restoration, was in charge of this treatment.
The painting, which represents the Adoration of the Magi, is a
seventeenth-century oil-on-wood panel attributed to a member of the very
productive Francken family of Antwerp. The support consists of three thin
(8 mm) oak boards, sawn on the false quarter and held together with pins.
The panel measures 71.4 104 cm. The seal of the guild of Antwerp, a
castle and hands, is stamped on the reverse.
Straightening of a
Distorted Panel
Method
366 Gl at i gny
The panel comes from the museum of the city of Ath, located
in a large eighteenth-century house whose rooms are damp and barely
heated in winter. There is no RH-control system. The harsh climatic con-
ditions had caused serious damage to the support. The joints of the panel,
blocked in its frame, had come apart, and several cracks had appeared
(Fig. 1). The boards presented a severely convex prole, in addition to a
spiral distortion. Tunnels of xylophagous insects had caused the wood to
become more reactive to variations in RH.
Except for an old restoration consisting of glued strips of linen
over the open joints, the support had never been altered. The paint layer,
however, was coated with numerous overpaintings. To mask irregulari-
ties in the support, the joints and crack areas had been broadly lled in
and retouched.
The distortions were disturbing to the viewer and made it impos-
sible to frame and display the work. Therefore, a decision was made to
straighten it with a backing while leaving the Antwerp seal visible. The
treatment of the support consisted rst of eliminating the linen reinforce-
ments, then consolidating the worm-eaten wood with a solution of 10%
Paraloid B72 in paraxylene, and nally of gluing the splits and joints
(Fig. 2). All the cavities were lled with oak sawdust sifted to less than
0.25 mm in a 25% polyvinyl acetate and water emulsion. Subsequently
the paint layer was protected by a facing of silk paper glued with beeswax,
and the panel was placed in a microclimate box, where the humidity was
367 Bacxi os or Pai rtn Patis . Rti rorctxtr an Cosrrai r
Fi gure 1
Francken family (attrib.), Adoration of the
Magi, seventeenth century. Oil on panel,
71.4 104 cm. Muse Athois, Ath, Belgium.
The condition of the painting before conser-
vation, with splits and cracks, is shown.
gradually increased. At 75% RH the panel was practically at; it showed
some remaining spiral distortion but had gained good exibility (Fig. 3). At
this juncture it was kept at in a room where the RH had been stabilized
at 75%. A layer of beeswax and damar resin (seven parts to two) was
spread over its surface. The wax-resin was applied with a warm brush and
smoothed out with a heating spatula. After it cooled, the rst 1 cm end-
grain layer of balsa wood was placed diagonally across the surface; then
the second was placed, overlapping the rst (Fig. 4). Finally, the excess
wax-resin was wiped o after it had been heated with warm air, and
bamboo-ber paper was glued with the same adhesive. Oversized balsa
blocks were sawn to t the panel. The Antwerp seal was made visible
again when an opening was cut out of the backing (Fig. 5).
The panel was gradually brought back to 50% RH. Once the fac-
ing was removed, the painted surface was cleaned and retouched (Fig. 6).
The painting was xed in its frame with springs and returned to the
museum in Ath. In the year since, no distortion has been observed. The
balsa-wood backing provides the work with good support and excellent
protection from that environments signicant uctuations of RH.
368 Gl at i gny
Fi gure 2
Francken family (attrib.), Adoration of the
Magi. The panel is shown after the gluing of
the splits and joints, and before straightening.
Fi gure 3
Francken family (attrib.), Adoration of the
Magi, in a microclimate box. The panel has
been considerably straightened.
369 Bacxi os or Pai rtn Patis . Rti rorctxtr an Cosrrai r
Fi gure 4
Francken family (attrib.), Adoration of the
Magi, reverse. A second layer of balsa blocks
is placed on top of the rst layer.
Fi gure 5
Francken family (attrib.), Adoration of the
Magi, reverse, detail. An opening cut in the
backing allows viewing of the Antwerp seal. Fi gure 6
Francken family (attrib.), Adoration of the
Magi, after conservation.
1 Patrick Mandron, from the Service de Restauration des Muses Nationaux; Aubert Grard,
from the Centre Rgional de Restauration et de Conservation des Oeuvres dArt; and Jean-
Albert Glatigny, from the Institut Royal du Patrimoine Artistique backed two paintings on
limewood, using the balsa-wood/wax-resin treatment. These works, which are by Lucas
Cranach the Elder, represent Saint Peter and Saint Paul, and they belong to the Louvre
Museum.
Beardsley, B. H.
1978 A exible balsa back for the stabilization of a Botticelli painting. In Conservation of
Wood in Painting and the Decorative Arts: Preprints of the Contributions to the Oxford
Congress, 1723 September 1978, ed. N. S. Brommelle, Anne Moncrie, and Perry Smith.
London: International Institute for the Conservation of Historic and Artistic Works.
Buck, R. D.
1970 The dimensional stabilization of the wood supports of panel painting. In Conference
on Conservation of Canvas and Panel Paintings, Warsaw, 1921 October 1970. N.p.
Habaru, S.
199091 Le doublage au balsa des peintures sur panneau. Diploma thesis, ENSAV
La Cambre, Brussels.
Mori, N.
199293 Les barrires contre lhumidit. Diploma thesis, ENSAV La Cambre, Brussels.
Spurlock, D.
1978 The application of balsa blocks as a stabilizing auxiliary for panel paintings. In
Conservation of Wood in Painting and the Decorative Arts: Preprints of the Contributions to
the Oxford Congress, 1723 September 1978, ed. N. S. Brommelle, Anne Moncrie, and
Perry Smith. London: International Institute for the Conservation of Historic and
Artistic Works.
vom Imho, H. C.
1978 Reinforcing a thin panel painting. In Conservation of Wood in Painting and the
Decorative Arts: Preprints of the Contributions to the Oxford Congress, 1723 September 1978,
ed. N. S. Brommelle, Anne Moncrie, and Perry Smith, 16568. London: International
Institute for the Conservation of Historic and Artistic Works.
References
Note
370 Gl at i gny
371
A
iarot rrororri o of panel paintings that have been thinned
and cradled exhibit damage caused by the cradle or signs of stress
from it. Environmental conditions play a large part in this equa-
tion. Much can be done by altering the environment to achieve stability,
even with the cradle left largely unaltered.
However, many panels are so stressed or damaged by the cradle
that it is essential to remove it. Some thinned panels are self-supporting
but vulnerable after removal from the cradle, basic consolidation, and
rejoining. Their response to environmental changes can be rapid and dam-
aging. In some cases an unattached auxiliary support can oer further pro-
tection and stabilitymore than that provided by careful framing and
tting of backboards. The auxiliary support allows reduced movement of
the panel within set limits. The panel is able to become alternately convex
and concave with changes in relative humidity (RH) while being retained
in the panel tray.
The reasoning of the cradle maker when thinning and tting a cradle to
a panel is as follows: The panel is thinned suciently to allow it to be
attened without immediate obvious damage occurring, and the cradle
is then glued in place. It holds the panel in a at plane while allowing
cross-grain expansion and contraction. The elements glued in the grain
direction are sometimes used to reinforce joins, damages, and splits while
retaining the sliding battens at suitable intervals. The sliding battens hold
the panel in a at plane and provide rigidity for the complete structure.
Several factors have been either disregarded or underrated in the
design and construction of cradles. For example, the inuence of the
glued members lying parallel to the grain should be considered, inasmuch
as the overlying areas of the panel are more stable, more rigid, less hygro-
scopic, and stronger than the unsupported areas; areas of adjacent stress
concentrations close to the glued members (Fig. 1)where the stress tran-
sitions are greatestcan show eects such as those seen in Figure 2a and
2b; the relative freedom of the unsupported areas between the glued
members allows them to react to stress and develop washboarding from
dierential movement movement (Figs. 2b, 3a), and the dierential caused
by unequal stresses can result in, or exacerbate, blistering and aking in
the ground and paint lm. It is important to note that all of these points
Cradling
A Flexible Unattached Auxiliary Support
Simon Bobak

372 Bobak
Fi gure 1
Representative areas of stress concentration
(marked by arrows) in a cradled panel.
a
b
Fi gure 2a, b
Damage caused by a cradle (a), and wash-
boarding caused by a cradle (b).
a
b
Fi gure 3a, b
Washboarding. This phenomenon is more
pronounced when the sliding battens are in
place (a); note the new camber when the slid-
ing battens are removed (b).
assume the correct functioning of the cradle. In practice, however, many
cradles lock up (because of inadequate clearance, poor construction, or
overgenerous use of glue in assembly), causing the type of damage typi-
cally exhibited by functioning cradled panelsalthough the damage is
often more severe.
Modifying the cradle
The type and amount of stress and damage from the factors discussed
above will determine the degree of intervention required. Decisions about
intervention are made on the basis of experience rather than of analysis. It
may be possible to remove the sliding battens from the cradle safely and,
by observing the change in curvature, make an assessment of the amount
of stress within the panel. The diculty of removal and the abrupt change
in curvature can make this a hazardous procedure, one requiring consider-
able care and experience. After the removal of the battens, it is important
to monitor the movement of the panel through several cycles of low and
high RH. It may take some days, depending on the RH required, for the
panel to reach an initial equilibrium, such as that shown in Figure 3b.
Some thinned, cradled panels show no signs of obvious damage.
Even if it has caused washboarding, a cradle may not appear to have pro-
moted further damage. It may be sucient (and, indeed, prudent) to
ensure the free movement and function of the cradle by one of the follow-
ing means: removing the sliding battens and sanding them to achieve a
looser t; reducing the thickness of the battens even further to increase
their exibility, a technique that allows the cradle and panel to achieve a
degree of curvature, as seen in Figure 4ad (the relationship between
373 A Fitxi tit Uarracntn Auxi ii ar Surrorr
a
b
c
d
Fi gure 4ad
Side elevations and end views of sliding bat-
tens and cradled panels showing reduction in
thickness by four methods: (a) the battens
relieved on the back face at their tips, a tech-
nique that increases cradle exibility at the
outer edges; (b) the battens relieved on the
front face at their tips, a technique that
increases their exibility and allows an imme-
diate unrestrained increase in panel curvature
at the outer edges; (c) the battens slightly
reduced in thickness, a technique that ensures
their free movement and the basic functioning
of the cradle, with very little increase in cradle
exibility; (d) the battens signicantly reduced
in thickness, and the consequent gaps in the
glued members lled with packing spacers, a
technique that increases cradle exibility.
thickness and exibility is fully explained in Marchant, Development of a
Flexible Attached Auxiliary Support, herein);
1
or applying self-adhesive
Teon PTFE tape to the sliding battens to reduce friction.
2
If needed, in
addition to easing the cradle, other possible improvements include con-
struction of shaped slips for the frame that follow the panels prole;
assurance of adequate retention in the frame even while it accommodates
some change in curvature without excessive restriction; tting of back-
boards, which also oers additional physical protection and slows the rate
of moisture exchange; provision of microenvironments (such as microcli-
mate boxes, glazing, and backboards) to stabilize the panel; and control of
the environment in the room or display area by attention to heat sources
and local hot spots (such as res and picture lights), drafts, and proximity
to windows, outside walls, and direct sunlight. While in many cases these
measures may be sucient, some panels are so stressed or damaged that
complete removal of the cradle is essential.
Cradle removal
Although this article does not propose to cover cradle removal in detail,
the following points should be considered: humidifying the panel before
cradle removal can reduce any sudden changes in curvature;
3
making a
bed that follows the panels curvature and irregularities and safely contains
the panel is vital;
4
and determining in advance the progression or direction
of removal is important, as it is possible inadvertently to increase the stress
locally while reducing it in another area.
The reaction of the panel to cradle removal may be discernible as
having several stages. The removal of the sliding battens can often trigger
an immediate increase in curvature. The removal of the glued members
down to a veneer thickness may not alter the curvature further. The
removal of the remaining veneer and animal glue can sometimes cause
the panel to increase its curvature (Fig. 5af ), although it may occasionally
decrease the curvature.
After cradle removal, rejoining, and consolidation of any dam-
aged areas, the panel may be self-supporting, although fragile and
dicult to handle safely. In that case, an unattached auxiliary support
may be considered as an option to support and protect the panel while
allowing it movement.
Size and thickness
In practice it has been found that panels larger than approximately 1 m
75 cm are not easy to accommodate using this system. Either they are of
such thickness that they do not require an unattached auxiliary support or
they are too thin, their strength-to-weight ratio being such that an attached
auxiliary support is required.
In addition, with a very thin, large panel, the weight alone can
trap the bottom edge and reduce the panels ability to move with changes
in humidity. This may lead to damage, even when Teon PTFE is used to
line the frame/tray rabbet.
Type of wood
There are great variations in the rate and amount of movement among
dierent types of wood, even aside from variations resulting from the cut
Suitability of a Panel for
an Unattached Auxiliary
Support
374 Bobak
of the timber, irregularities, or damage. No denitive rules can be fol-
lowed, but the amount of movement a panel is expected to make must be
considered. The depth of the frame or tray required will need to be con-
sidered if the curvature is expected to be great.
Grain orientation
Preferably, the grain of a painting panel should be vertical, because the
endgrain is less prone to accidental damage and compression and is best at
load bearing. When the grain is vertical, the exible battens also function
more easily because the bottom edge is less likely to be trapped, as there is
no change in the angle of the bottom of the panel in relation to the trays
bottom rabbet. However, a panel with a horizontal grain direction can still
be accommodated by the tray and exible support if attention is paid to
the weight of the panel and to its bottom bearing edge with regard to its
frictional resistance to movement.
Panel condition
In assessing the condition of a painting panel, several points should be
considered, including worm damage; areas of sapwood; timber decay;
cracks; checks; repaired splits; original joins and rejoins; buttons,
5
insets,
butteries,
6
and other kinds of repairs; and any other previous conserva-
tion work. All of these can aect the strength and modulus of elasticity
7
of
the panel and must be considered in any assessment. No conclusive advice
can be given; however, the conservator must be condent that the panel is
strong enough to deect the auxiliary support safely.
The exible unattached auxiliary support, described herein as a
system for retaining and supporting vulnerable panels while allowing con-
vex and concave movement, could also be considered as an alternative
(albeit a time consuming and complicated one) to conventional framing of
panels that do not require a support per se. In view of the number of pan-
els that are damaged by misconceived framing techniques, perhaps the
greater investment in time would be worthwhile.
375 A Fitxi tit Uarracntn Auxi ii ar Surrorr
a
b
c
d
e
f
Fi gure 5af
When the cradle is removed from a cradled
panel (a), the panel (shown in ve sections)
assumes typical curvature proles (bd) as it
moves toward an initial equilibrium (f ).
The exible auxiliary support and tray, in addition to retaining the panel,
damp
8
its movement by applying a measured restraint while allowing con-
vex and concave (or reduced-curvature) movement. The exible battens
accommodate an increase in curvature while encouraging a return to the
neutral position against the shaped prole of the panel tray. The back
spring accommodates concave or reduced curvature and also encourages
a return to the neutral position toward the panel tray prole.
The assembly is composed of several parts: the exible batten, the
back spring, the central bearing, and the bases (Figs. 69).
Flexible batten
Flexible battens, if correctly rated for exibility to the panel, are able to
be deected by the panel. This allows for an increase in curvature when
the RH drops (Fig. 10). The interdependency of the exible batten and the
back spring should be noted. The exibility of the batten is increased
toward the tip by one of the following methods:
Tapering of thickness
This method involves the reduction in the thickness of the batten by thin-
ning toward the tip from the center. This reduces the stiness toward the
tip, thus alleviating the problem of all the loads being referred inward
toward the central area of the panel and, consequently, increasing the
moment
9
toward the central axis.
Tapering of width
This method (as outlined in Marchant, Development of a Flexible
Attached Auxiliary Support, herein) is also suitable and may be preferred
because graduating the exibility is more accurate and more easily
achieved. (The number and spacing of the exible battens are covered in
Principle of the Auxiliary
Spring System
376 Bobak
Tray
Flexible
batten
Panel
Back
spring
Central
bearing
Base
Backboard
Flexible
support
spring
assembly
Tray
b a
Panel
Flexible
support
spring
assembly
Backboard
Fi gure 6a, b
Complete system (a) consisting of a tray,
panel, exible support spring assembly, and
backboard; the same components are shown
disassembled (b).
the section entitled Matching the Support to the Panel, below.) In all
cases the exibility of the battens and back springs combined must be
greater than that of the panel in order to ensure that the support yields
to the panel.
Back spring
The exibility of the back spring will determine the preload that keeps the
panel in position against the frame rabbet and the ability of the panel to
become concave or to decrease its curvature (Fig. 11). The exibility of
the back spring can be varied by increasing or reducing its width, b,
increasing or reducing its thickness, d, or altering its span. These three fac-
tors can be adjusted according to the panels size, weight, and curvature.
It should be clearly understood that the deection of the back
spring is in a constant ratio to load, and most of the forces generated by
the panel will be referred to the central area of the panel parallel to the
grain if it becomes concave or reduces its curvature from the neutral slip
shape. Thus, of the two exible parts, the back spring can have the more
critical inuence. However, this fact should be balanced by the knowledge
that most problems occur when panels are restrained from becoming con-
vex (viewed from the front) rather than from becoming concave.
377 A Fitxi tit Uarracntn Auxi ii ar Surrorr
Fi gure 7, above
Flexible support spring assemblies on a
backboard.
Fi gure 8, above ri ght
A panel in place on exible support spring
assemblies.
Fi gure 9, ri ght
Model of the spring support assembly with a
panel at a neutral curvature prole (55% RH).
Fi gure 10, bel ow
Model of the spring support assembly with a
panel at an increased curvature (low RH) in
comparison to that shown in Figure 9.
Fi gure 11, bel ow ri ght
Model of the spring support assembly with a
panel at a decreased curvature (high RH) in
comparison to that shown in Figure 9.
Central bearing
The thickness and area of the central bearing is determined by the maxi-
mum curvature expected (Fig. 12a), the available depth within the frame
and tray, and the desire to make the bearing as short as possible to reduce
the hard point in the center of the exible batten and the back spring.
Bases
These form a bridge enabling the back spring to function. Their area and
depth are guided by the same factors as the central bearing. The bases are
glued at the ends of the back spring and consequently will reduce the
eective span of the back spring by their length.
Only one of the two bases should be xed to the backboard. If
both were glued, then a rigid arch would be formed, and the exibility of
the back spring would be greatly reduced. The maximum height of the
bases must allow for the further expected deection of the back spring. In
more recent developments, Plastazote foam,
10
sandwiched between timber,
has been used in the bases to allow movement in the back spring when
both bases are xed to the backboard (Fig. 12a, b).
By taking proles at frequent intervals with the RH constant at 55%,
11
it is
possible to monitor the curvature of the panel and record its prole when
it has reached equilibrium. This curvature, if any, is taken to be the neu-
tral position. When there is regular slight curvature, it is not necessary to
shape the battens to the panel, since a small preload is desirable to keep
the panel in position against the tray rabbet and to bias the panels move-
ment in the preferred direction.
Where the curvature is large or uneven, the batten will need pack-
ing to the surface prole of the back of the panel. This is done by the use
of short, shaped sections of balsa wood glued to the front of the batten.
The balsa grain running at 90 to the batten grain will minimize any
change in exibility (Fig. 13).
Safe deection of the panel
In order to establish safe deection, test samples and model panels can be
made. Although they must not be relied upon to give analytical informa-
Matching the Support
to the Panel
378 Bobak
a
b
Plastazote foam
Plastazote foam
Fi gure 12a, b
Critical clearance points (marked by arrows)
for maximum (a) and minimum (b) curvature.
tion, they are nevertheless helpful in establishing the broad range of likely
forces. Test samples can never represent the exact structure of the panel,
its paint and ground layers, or the weaknesses of irregularities and aging.
Gently exing the panel can help to verify sample data and must be done
with the greatest possible prudence. A rig using a spring balance or small
weights can also relate load to movement.
Once the total safe load for the panel is established (including a
safety margin to ensure that the support will yield to the panel), the safe
load is divided by the number of elements in the support to nd the load
per element, which produces a determined deection. The exible batten
is then thinned to give the determined deection at that load. The number
of elements in the support will be determined by the area of the panel and
the length of the panel along the grain. In practice, most panels have
spring elements with centers between 100 mm and 150 mm. The back
spring must be just sti enough to ensure that the exible batten is held
fully engaged, in contact with the back of the panel, thus providing an
even support. It must not be so sti that the concave movement of the
panel is restricted.
Eect of batten curvature on panel curvature
To summarize, if the neutral curvature prole of the exible batten is less
than that of the panel, then the panel will be moved toward a atter plane.
If the neutral curvature prole of the exible batten and that of the panel
are the same, then the panel will have no tendency toward either concave
or convex movement. Thus it is possible to tailor the spring system to
encourage a panel toward a atter plane.
The timber used is Sitka spruce (Picea sitchensis). It is straight grained,
largely free of faults, light, elastic, and of consistent density. It has been
used in aircraft construction for more than ninety years and is available
with aircraft release notes.
12
The face of the exible battens touching
the verso of the panel can be covered with felt or cotton tapes to protect
the panel from abrasion. The panel tray is constructed from hardwood
(mahogany or a similar wood) to achieve rigidity for a minimum size of
section. Saw kerfs
13
can be cut in the mitered corners and hardwood
tongues glued in place to increase rigidity (Fig. 14). This method is espe-
cially good for small trays when the timber section is insubstantial. In
some cases there may be insucient space for a tray within an existing
frame; it may be possible to use the frame as the basis for the shaped slip
pieces and to build up the sides of the rabbet so as to make the frame
eectively into the tray. The backboard and spring supports may be tted
by the same method.
All four sides of the panel have their proles taken, and these are
transferred onto the tray edging section. It is generally easier to construct
the shaped proles and glue them into a basic section than to carve them
Construction
379 A Fitxi tit Uarracntn Auxi ii ar Surrorr
Fi gure 13
Part of a exible batten showing cross-grain
balsa-wood packing pieces shaped to t the
back of an irregular panel.
out of the solid wood. Many panels have a propeller-like twist in addition
to any curvature. Such twisting must be carefully considered when estab-
lishing datums for construction within the tray in order to balance the
diagonal distortion evenly.
All visible edges of the tray can be toned, gilded, and distressed
to match the existing frame. Frequently the sight size of the frame is too
large, and consequently part of the trays edge can be made to project
beyond the rabbet and become visible as an inner slip.
The backboard is made of marine-quality plywood,
14
which may
be obtained with good-quality veneers. Other stable sheet materials may
be used, the choice being made on rigidity, thickness, weight, and appear-
ance. The backboard must be strong enough to withstand the loads
imposed by the support with little deection.
The panel is placed in the tray and retained by the spring sup-
ports and backboard. The backboard is then screwed into the tray edge
section. Finally, the completed assembly is tted into the frame with brass
strips and screws.
There are disadvantages to the exible unattached auxiliary support sys-
tem: it is only suitable for a limited range of panels; the assessment of
forces and panel strength is largely empiric; some panels and frames will
not accept a deep tray without the result appearing ungainly; and it is not
possible to see the back of the panel without the removal of the back-
board and support. Despite these limitations, there are several advantages
to the system. For example, there is a minimum of interference with the
original panel; concave and convex movement of the panel is possible
without overstressing; known forces are applied to the panel; the panels
movement within the tray indicates changes in RH and alerts conservators
to inadequate RH control; RH changes are buered by the tray and back-
board; and physical protection of the panel is providedan especially
important consideration when the panel is out of the frame.
1 I bd
3
12, where: I moment of inertia; b breadth (width); and d depth (thickness).
With a constant thickness, d, if the width, b, is halved, the deection for the same load will
double. If d is halved, the deection will increase by eight times with the same load (see
Marchant, Development of a Flexible Attached Auxiliary Support, herein).
2 Teon PTFE (polytetrauoroethylene) is a tape with very good properties for reducing friction.
3 To humidify a panel, the RH in the room may be increased to 6570% for several days to
reduce the stress within the panel if the convex curvature is expected to be large.
Notes
Assessment of the
Unattached Flexible
Auxiliary Support System
380 Bobak
Fi gure 14
The corner of a panel tray showing saw kerfs
in which hardwood veneers have been
inserted, to increase rigidity and the strength
of the joint.
4 A bed may be made from a prepared board that is packed with balsa wood strips of varying
size and thickness to support the panel over its entire surface during cradle removal. While the
cradle is being removed, it will need to be adjusted continually if the curvature alters.
5 Buttons, also known as cleats, are the rectangular reinforcing blocks frequently glued over
the back of the panel to repair cracks. While they are generally cut into the surface, they may
be left proud.
6 Butteries are the bow-tie or buttery-shaped repair blocks often cut into the back of the
panel to reinforce cracks, splits, and joins.
7 The modulus of elasticity is constant for a particular material. It is the force above which the
panel will deform or be damaged and not return to its original condition by elastic behavior.
8 To damp the movement of the panel is to reduce the amplitude of the cycles.
9 The moment is the product of the force and the distance from its point of action.
10 Plastazote foam is a closed-cell, cross-linked polyethylene foam (see Materials and Suppliers).
11 In the United Kingdom, 55% RH is generally considered the best average humidity in which to
keep panel paintings.
12 In the United Kingdom, aircraft release notes identify timber that is tested to Civil Aviation
Authority standards, for consistency of density, quality, and moisture content.
13 Slits made by a saw blade.
14 British Standard 1088 signies marine quality, indicating that the stability and quality of con-
struction are assured by testing. It is not the same as waterproof plywood (WPB), which is
produced to a lower standard.
Plastazote foam (REF LD 24), BXL Plastics, Mitcham Road, Croydon, Surrey CR9 3AL, U.K.
(distributed by Hemisphere Rubber Co., 65 Fairview Road, Norbury, London SW16 5PX, U.K.).
Teon PTFE, CHR Industries, 407 East Street, New Haven, CT 06509. European supplier: Furon
CHR Products, P.O. Box 124, 7640 AC Wierden, Netherlands. United Kingdom distributor:
Polypenco, now part of DSM Engineering Plastic Products UK, 83 Bridge Road East, Welwyn
Garden City, Hertfordshire AL7 1LA, U.K. (The PTFE tape is marketed as Temp-r-tape HM series.)
Materials and Suppliers
381 A Fitxi tit Uarracntn Auxi ii ar Surrorr
M
osr rati xouri o and display requirements can be
achieved with conventional methods of framing and retention,
but occasionally a dicult problem will arise in which recog-
nized methods of support are inadequate. This article documents the
development of an alternative approach to some of the more dicult
problems encountered in the support of weak or responsive panels.
Description
One such problem occurred in 1989 with the conservation of a large
sixteenth-century Flemish panel. The painting, measuring 1.2 1.7 m,
consisted of six oak boards joined horizontally. It had been thinned to
between 6 mm and 8 mm and had a late-nineteenth-century heavy pine
cradle attached, constructed from nine xed horizontal members, each
measuring 50 mm wide 25 mm thick, and six vertical sliding battens,
each 60 mm wide 15 mm thick.
As with so many cradles of this period, the device exhibited good
workmanship but was intended to atten the panel. Because the cradle
was of a rigid construction with minimum tolerances allowed for move-
ment of the sliding battens, it was potentially damaging. Subsequent to
the cradles installation, variations in environmental conditions caused the
panels moisture content, and hence its curvature, to alter. The cradle
could only accommodate a small change in the panels warp before the
battens became locked, preventing further movement. Stresses then devel-
oped, causing fracturing and partial disjoins to occur in a number of places
on the panel. An assessment of the condition of the painted surface
showed that many of the structural faults had produced corresponding
damages to the ground and paint layers.
When the panel arrived for treatment, its prole viewed from
the front was concave, and the cradle was totally seized. The panel paint-
ing was in very poor structural condition and the concern was that it
would deteriorate further. It was considered that just freeing the sliding
elements of the cradle would not provide an adequate solution to many
of the problems. Therefore, it was decided that removal of the cradle
was necessary to complete the repairs satisfactorily, as well as to ensure
future stability.
First Case Study
382
Raymond Marchant
The Development of a Flexible Attached
Auxiliary Support
Structural condition
When the cradle was removed, the extent of the weaknesses and damage
to the panel could be fully appreciated. One fracture, which ran almost the
whole length of the panel, had occurred in an area of worm-damaged sap-
wood adjacent to a join. The fracture was so severe that one of the board
sections was virtually hinged to the main body of the panel only by a
number of small areas of intact bers.
There were also traces of two 10 cm wide cross-grain channels
across the panel that could just be discerned in the thinned surface. These
traces indicated that battens may have been present before the cradle was
tted. Exposed dowels showed that the panels original thickness had been
reduced by about half when it was thinned. It was likely that this panel
had had a history of structural problems long before the cradle was tted.
After structural repairs, rejoins, and consolidation had been car-
ried out, the panels cross-grain prole was monitored and recorded sev-
eral times during a period when the relative humidity (RH) was allowed to
vary widely. To judge the panels response to likely extremes of environ-
mental conditions, its prole was recorded at 40%, 55%, and 75% RH, and
its condition was reassessed. Monitoring was carried out with the panel
standing vertically on its endgrain.
Released from the cradle, the panels prole altered considerably,
becoming convex when viewed from the front and responding quickly to
even small changes in RH. Because of its thinness, the strength-to-weight
ratio, although improved by repairs, was so poor that it could be handled
only with great care. If laid horizontally, it was subject to the risk of frac-
ture if any attempt had been made to lift it by one of the long-grain edges.
Another cause for concern was an area of severe worm damage,
again in a band of sapwood extending across the board, close to the bot-
tom, supporting edge. This weak edge would be subject to damaging
forces imposed by the weight of the panel bearing on it and the need for it
to move to accommodate changes of curvature. If this natural tendency to
warp were again restricted by a rigid secondary support, further damage
caused by compression and/or tension perpendicular to the grainwould
be likely to occur.
Identifying the need for an attached auxiliary support
It was apparent that if the panel were to remain stable without suering
further damage, a method of support other than those normally used was
needed. Because the problems presented by this panel were known to be
dicult to resolve satisfactorily, it was decided that the options should
be considered very carefully before a course of action was decided upon.
Conventional techniques of support and retention of RH-responsive
panels include sprung-metal clips, secured within a frame rabbet; a foam-
cushioned panel tray support (Brough and Dunkerton 1984); and un-
attached auxiliary exible supports (see Bobak, A Flexible Unattached
Auxiliary Support, herein).
In these examples of unattached supports we nd a common prin-
ciple: retainers exert pressure on the back of the panel, and this pressure
frequently concentrated around the perimeter or on the line of the central
long-grain axisis balanced by the reaction of the lip of the frame rabbet
acting against the edge of the face of the panel.
383 Tnt Dtvtiorxtr or a Fitxi tit Arracntn Auxi ii ar Surrorr
For a self-supporting panel, a shaped slip would normally be made
to suit the panel prole when the panel is stabilized at 55% RH. If environ-
mental conditions remain stable, good contact should be maintained with
the shaped slip on all four edges, and only small, balanced reaction forces
will result (Fig. 1a). Problems can arise, however, as soon as conditions
change (Fig. 1b, c).
Figure 1 shows a panels response to environmental changes.
Dierential absorption or loss of moisture content in the panel, due to
changes in RH, cause it to warp (Thomson 1978:20810). The opposing
forces illustrated in Figure 1b and 1c may result in bending stresses, which
in an already weak panel could result in fracture. These adverse eects are
further accentuated when the grain runs horizontally, because the weight
of the panel resting on its supporting edge causes frictional resistance to
the movement needed to accommodate a change in curvature. In large
panels, forces can be magnied by leverage to produce dangerously high
concentrations of stress some distance from where the resistance to move-
ment occurs. If an area of weakness exists, failure is likely to occur there.
Under these circumstances, the use of one of these types of secondary
supports would not be satisfactory.
Having fully assessed the condition of a panel, the panel conserva-
tor must make a decision as to whether an attached support will be neces-
sary. After removal of a cradle or damaging support from a panel, it would
be preferable not to have to make any further attachment. However, there
are circumstances in which this measure cannot be avoided.
As a general rule, if an unframed panel cannot be handled
condently or will not safely support its own weight when placed horizon-
tally on a surface, then an attached support should be considered in order
to provide the required reinforcement. It is almost impossible to reinforce a
weak panel without using an attached support. But an attached support can
be designed to ensure that it is in sympathy with the panels requirements.
A reinforcing structure is required to help strengthen a weak panel
and assist in spreading stresses more uniformly. The other function of a
secondary support in this situation should be to act as a restraint by allow-
ing changes of curvature to take place in a controlled manner and within
predetermined limits. Therefore, the secondary support should be exible.
This design concept was successfully established by Simon Bobak (see A
Flexible Unattached Auxiliary Support, herein) for the unattached support
of small panels but would need considerable development before it could
be applied to an attached support for large, heavy panels.
Batten design
In an attempt to design an attached support that would fulll these basic
requirements, an analysis was rst made of the eect of attaching a uni-
form rectangular-section batten to a curved surface. It was hoped that this
would also provide a better understanding of why some cradles, even if
they allow movement, still have a damaging eect on panels.
The simplied representations in Figure 2ad show curves
achieved by loading a uniform rectangular-section batten. The curvature of
the panel is shown exaggerated as an arc with a constant radius of curva-
ture (Figs. 2a, b). The batten is deected within the arc by the application
of a force at its center (Fig. 2b). This situation may also be represented dia-
grammatically as a simply supported beam loaded at its center (Fig. 2d).
384 Marchant
55% RH

a
Below 55% RH

b
Above 55% RH

c
Fi gure 1ac
A panel retained against a slip prole using
spring framing clips. The magnitude of pres-
sure at these points and the reaction of the
proled support is shown by the arrows. At
55% RH (a), minimum pressure is applied at
four retaining points. Below and above 55%
RH (b, c), the magnitude of pressure increases
and the reaction of the proled support is
concentrated at the contact points. These
forces would be similar if other unattached
support systems were used as a method of
retention.
If the batten is deected within the arc by a force at its center,
the only point of contact with the arc other than the outer edges will be
at the center. The deection curve in Figure 2b will be the same as that
represented in Figure 2d. It will not have an equal radius of curvature
over its length but will be straighter toward its ends in the form of a
parabolic curve.
To produce contact with the arc at points toward the ends of the
batten, greater force would be required at those points to make the batten
deect. If the uniform rectangular-section batten were to be attached at
a number of points to the curved surface, as in Figure 2b, it would have a
greater straightening eect on the surface (inducing greater tension at the
attachment points) toward the outer edges. To avoid the problem of creat-
ing high stress toward the edges of panels (which occurs with many con-
ventional cradles), the battens should be made progressively weaker
toward the ends.
Shape and section
Ideally, therefore, a batten is needed that would have an equal straighten-
ing eect at all points along its length. To produce a batten that will bend
with a constant radius of curvature under the conditions outlined, it is
useful to understand some basic structural theory. The relationship
between stress and curvature of a member when subjected to a simple
bending moment is given by the equation:
where: M is the bending moment (a function of load and distance); I is the
moment of inertia of the section (a function of breadth and depth); E is
the modulus of elasticity
1
of the material (a constant); and R is the radius
of curvature.
Therefore, for R to be constant along the length of the batten,
EI/M must also be constant. As M decreases linearly away from the center
toward the ends and E is not variable, then I must decrease in the same
385 Tnt Dtvtiorxtr or a Fitxi tit Arracntn Auxi ii ar Surrorr
a b
c
d
w
w
w
2
w
2
Fi gure 2ad
A uniform rectangular-section batten
deected within an arc by a force at its center.
A section of panel is shown (a) with constant
radius of curvature and a straight uniform-
section exible batten. The batten deected
by a force (W) at its center (b) contacts the
curved surface only at the center and the
outer edges. A batten supported horizontally
at its center (c) deects in the same way as in
b and d when supporting the weight of a
curved panel. A batten simply supported at its
ends and loaded at its center (d) deects into a
parabolic curve.
ratio as M. As I bd
3
12, either the breadth, b, or the depth, d, could be
chosen as the variable factor to produce the linear decrease.
The breadth of a rectangular-section member is directly propor-
tional to its deectionthat is, if the breadth, b, is doubled, then twice the
load is required to produce the same deection. But if the depth, d, or
thickness, is varied, the stiness will alter as the cube of d. That is, if the
thickness is doubled, then eight times the load needs to be applied to pro-
duce the same deection, or if the thickness were halved, then under the
same applied load, the deection would increase eight times.
It follows that it would be dicult to produce the linear decrease
required if thickness were chosen as the variable factor. The resulting bat-
ten would have a complex curved prole that would be dicult to deter-
mine and to execute accurately (Fig. 3a).
The alternative is to vary the width. Simply reducing the width at
a constant rate from the center toward the end satises the conditions for
producing a conguration of section which will deect into the uniform
curve required (Fig. 3b).
This shape of section is easy to produce. Its exibility can be
increased simply by reducing its thickness, and because it is a at section,
it is easy to incorporate into a support system. If this tapered batten is
now brought into contact with a curved surface until it deects, it will
conform more closely to the surface prole. If a number of attachment
points are made so that the batten has a straightening eect on the curved
surface, the tension at those points will be more equally spread, producing
an even restraint.
If calculations are made for deection based on a uniform rectan-
gular section, which then has its width tapered, the deection will increase
by about 50%. Allowance can be made for this. It is preferable, however, to
err on the side of exibility. An excessively sti support may damage the
panel, but problems are unlikely to occur if the support is too exible. It
should be able to yield to the bending force exerted against it by the panel.
To achieve reliable results from calculations, a suitable timber
needs to be specied. The timber chosen for the lattice components was
Sitka spruce,
2
which has excellent properties for this type of application. It
can be obtained in large, straight-grained, knot-free sections. It is also light
but strong, with consistent characteristics of exibility (i.e., E values).
386 Marchant
Side elevation Side elevation
Isometric view Isometric view
Plan view Plan view
a b
Fi gure 3a, b
Two congurations of batten shapes that
deect with a uniform radius of curvature.
Calculating batten flexibility
To calculate the required exibility of a batten for restraint, it is necessary
to know what bending force will be exerted against it by the panel. When
environmental conditions alter, moisture transference in the panel structure
generates internal forces. This bending force will produce pressure against
anything that restrains the panel from changing its curvature. It is possible
to measure empirically how much resistance is necessary to counteract this
change, but with a fragile panel, there is the risk that it may fracture before
any relevant information is obtained. It is not possible to predict the resis-
tance to bending that a weak panel will withstand before it fails; therefore,
some other means of assessing a loading gure for the batten needs to be
found. This can be done by considering reinforcement rather than restraint.
For simplicity, the calculation example that follows is based on a
batten supported at its center treated as a cantilever, with a fraction of the
panel weight used as the load gure (Fig. 4). (This concept will be
explained more fully in the section below entitled Evaluation of batten
exibility.)
For a cantilever, the deection () at the end under a single point
load is given by the equation:
where: deection; W load; L length of cantilever; E modulus
of elasticity;
3
and I moment of inertia.
Example. The following is a calculation of the thickness of the
battens that will support the weight of a panel horizontally within a
known deection. All other factors have been specied, including the
number, length, and width of the battens and what is considered to be a
safe limit of deection of the panel.
Deection () 30 mm
Panel weight 22 kg
Number of battens 10 Load at each end of each batten 1.1 kg
Therefore, W 1.1 9.80665 10.787
Length of batten 1200 mm
Cantilever length (L) 600 mm
Width of batten (b) 50 mm
Modulus of elasticity for Sitka spruce,
E 11100.6 n mm
2
387 Tnt Dtvtiorxtr or a Fitxi tit Arracntn Auxi ii ar Surrorr
L
W
D
e
f
l
e
c
t
i
o
n
Fi gure 4
Diagram of a cantilever deected by point
load at end. W load; L length of
cantilever.
The results of calculations are easily veried using prepared
sample battens and weights. It is not suggested that support battens be
specied purely by theoretical calculations but rather that calculations may
serve as a useful shortcut to produce sample sections for empiric evalua-
tion. It then becomes a question of judgment based on experience to
decide whether, or by how much, to alter such a batten to suit the particu-
lar requirement.
It should also be stressed that even for those with no understand-
ing of structural design theory, there is at least one important relationship
included in the equations that should be recognized. This is the correlation
between section thickness and exibility (as discussed above). In the design
of a secondary support, or even in the thinning of battens to ease an exist-
ing cradle, the result of reducing thickness by what may appear to be only
a small amount can have a very dramatic eect on the exibility of the
support. Conversely, it is very easy to produce an auxiliary support many
times more rigid than is necessary to perform its functionwith a conse-
quent risk of damaging the panel.
Method of attachment
With the form that the exible battens should take having been established
in principle, the next problem to consider was the method of attachment
to the panel.
The main factors to consider were as follows: It should not be
possible for the battens to seize, thus restricting dimensional changes in
the panel. The attachment of retaining points to the panel should be
achieved without the creation of rigid glue areas that are larger than nec-
essary or that extend too far across the grain, as this could contribute to
the characteristic washboard eect and the tendency to fracture at the
transition edges of glue areas. And it would be an advantage if the means
of attachment allowed for removal of the battens.
All of these basic requirements were achieved by the use of
exible retaining strips against the surface of the battens, held in place
parallel to the panel grain with slotted retaining blocks glued to the panel.
The blocks were made narrow in the cross-grain direction, and their size
was limited according to the number used. The greater the number of
blocks, the less tension each had to bear individually and the smaller the
glue area needed for safe attachment. For compatibility with the panel,
the blocks were made of oak. Evo-Stik polyvinyl acetate (PVA) wood-
working adhesive was used for the glue joins.
4
The number of retaining
strips, and hence the distribution of blocks, is determined by such factors
as the number of boards making up the panel,
5
surface irregularities that
may make attachment points dicult, areas of weakness that should be
avoided, and original features that one would prefer to leave unobstructed.
Using retaining strips against the face of the battens instead of
anchoring the battens directly to the panel ensured that there was little
risk of seizure occurring. However, it was also necessary to stop the indi-
vidual battens from moving and becoming misaligned. This was done by
linking them together in an accurately spaced conguration, with thin, at
timber strips used to create a lattice.
Finally, a supporting timber section was made to t under the bot-
tom edge of the panel. This skid strip was joined to the tips of the lattice.
It provided protection for the weak load-bearing edge, as well as providing
388 Marchant
a smooth, at surface to aid movement, reducing the risk that the panel
would stick in the frame rabbet or tray. Free movement was further
improved by using Teon/PTFE (polytetrauoroethylene) pressure-
sensitive adhesive tape
6
to line the rabbet.
Upon completion, the support lattice was attached to the panel
by engagement of the exible strips in position in the retaining blocks
(Fig. 5). When this procedure was done, the panel tended to atten out
slightly and, when handled, could be felt to be appreciably less exible
than before the auxiliary support was in place.
Monitoring panel warp
At this stage, the slip prole was considered. Assuming that enough time
has been available, the panel should preferably have had its end-grain
proles monitored and recorded three times during cycles of RHinitially
with whatever cradle or restriction was in place when the panel arrived for
treatment; again, with restrictions removed and the panel totally free to
respond; and, nally, with the new support attached. This prole would be
expected to fall somewhere between the rst two recorded proles.
Consideration should also be given to simulating the conditions
under which the panel is going to be displayed in the future. In some
countries the extremes of RH may be outside of the limits normally
used in a monitoring cycle (i.e., 4080% RH). After the slip prole has
been determined by monitoring under appropriate conditions
7
with the
support attached, some thought can be given to the depth of the tray or
rabbet. This depth needs to be sucient to accommodate the anticipated
extreme limits of movement of the panel; it should also be adequate for
the spring bridge supports, which will be used to hold the assembly in
place within the frame.
Back springs
The principle of using back springs was conceived by Simon Bobak (see A
Flexible Unattached Auxiliary Support, herein) for use on unattached sup-
ports. It consists of individual exible battens, each attached by a center
389 Tnt Dtvtiorxtr or a Fitxi tit Arracntn Auxi ii ar Surrorr
Fi gure 5
Back of sixteenth-century Flemish oak panel
(rst case study), 1.2 1.7 m, after structural
conservation, with exible auxiliary support
engaged, providing reinforcement.
pad to a spring bridging strip, with feet at each end for mounting on the
backboard (Fig. 6a).
This arrangement, which allows both increase and decrease of
curvature to take place in the panel while it maintains contact with the
support, was retained in principle but modied to suit the new lattice
design (Fig. 6b).
It was considered that one function of the action of the bridges
could be improved if they were inverted with both feet mounted onto the
battens, thus providing two reasonably spaced points of pressure against
the battens. This arrangement would encourage return movement
equally of the top and bottom of the panel to a neutral position, when
the curvature reduces, rather than the panel pivoting on the center pads.
In order that both feet could be mounted on a surface with variable cur-
vature, the timber pads were given a Plastazote
8
foam core, allowing
them to adjust to the changes. The pressure pad, which would now be in
contact with the backboard, was also made into a timber-foam sandwich
so as to prevent the creation of a rigid area being in the center of the
spring strip. The modication to the pads improves the overall cushioning
eect and allows dierential changes of curvature, dimension, and align-
ment to be absorbed.
Another advantage gained by inverting the bridges is that a nar-
row bar can be used to bear against the pressure pads. Previously, if a bar
were used, it would have had to be wide enough to engage both bridge
feet, or else a backboard would have had to be rigid enough to take the
spring pressure without bowing.
If a retaining bar is used to take the spring pressure, then the
backboard can be reduced in thickness and weight (which may be consid-
erable on a large panel) and can then act purely as a lightweight environ-
mental barrier (Fig. 6c illustrates this later development). The position of
the bar should be such that it engages to produce a slight preload of the
spring bridges just adequate to retain the panel against the slip prole.
(Note: Most pressure will occur against the bar during high RH, when the
panel will tend to atten, producing a far greater deection of the spring
bridges than when curvature increases.) In this particular case, the back-
board was a single sheet of plywood with a reinforcing section of timber
glued to the underside to stien it (Fig. 6b). In later supports, the improve-
ment of a rigid framing bar was adopted.
Evaluation of batten flexibility
Throughout this development, probably the most dicult judgment to
make was to determine the degree of stiness or exibility of the sup-
port lattice to match the panels requirements. With experience, it is pos-
sible to make a reasonable assessment of the strength of small panels,
but when a panel is so large that it cannot safely be lifted, handled, and
exed by one person, this becomes very dicult. Even when it is within
a manageable size, it is not easy to evaluate hidden weaknesses resulting
from small fractures, compression damage, and structural deterioration
resulting from age. E values (modulus of elasticity) cannot be used to
assess strength (resistance to bending) in the cross-grain direction. Tables
of E values for timber only apply to bending at points along an axis par-
allel to the grain.
9
390 Marchant
Panel weight as a factor in evaluation
During the development of this type of auxiliary support, the rst panel to
be assessed had lines of weakness caused by fractures and worm damage,
which made evaluation of its strength very dicult. Due to its areas of
weakness, the panel was assumed to have little or no inherent strength.
The intention of calculating a lattice exibility was to nd one that would
provide the reinforcement to support the weight of the panel horizontally
within a safe limit of deection.
The known factors upon which a judgment could be based for the
lattice exibility were the weight of the panel and the change in curvature,
monitored at the lower limit of RH that the panel might reasonably be
expected to be subjected to in the future, measured at the outer long-grain
edges of the panel as the dimensional deection from the center. The
panel weight, divided by twice the number of battens in the lattice, was
taken as the load that, when applied to one end of a batten, would produce
a similar deection from the center as that previously measured in the
panel. Tapered battens were then produced to give the specied exibility.
The result was that when the lattice assembly of battens was placed hori-
zontally on a central support and the panel placed on top, the panel weight
was adequately supported without the determined safe deection being
exceeded. The degree of rigidity of the support was therefore considered
correct for reinforcement.
When the support lattice was completed and anchored to the
panel, the assembly was evaluated in the vertical plane and found to give a
satisfactory degree of restraintit reduced the panels previous curvature by
about 30%. The panel could also be handled with much more condence. It
391 Tnt Dtvtiorxtr or a Fitxi tit Arracntn Auxi ii ar Surrorr
Rigid timber beam
used as retaining
bar to secure
assembly in frame
Backboard
thin plywood
Pressure pad
with foam core

Reinforcement
Mounting pads
with foam core
glued to batten
Backboard
reinforced with
section in center
Mounting pads
(one of each pair
glued to backboard)
Frame with slip profile

Panel

Flexible batten
Pressure pad
(central bearing)
Flexible bridging strip
Backboard
a b c
Fi gure 6ac
The main stages in the development of an
auxiliary support system. With an unattached
support (a), when the backboard is removed,
the spring bridges and battens joined to it
come away, leaving the panel loose in the
frame. With an attached support, in its earliest
development (b), spring bridges are inverted
and attached to battens. When the backboard
is removed, the panel and secondary support
remain loose in the frame. With an attached
support, in its later development (c), the panel
and secondary support are retained by a fram-
ing bar. When the backboard is removed, the
panel remains secure in the frame.
was not considered necessary to alter the lattice, and the project was com-
pleted by mounting the assembly in a tray with a spring-bridge support
behind the lattice. The overall result appeared to be perfectly adequate even
though the original design data were so limited.
This method of estimating the lattice exibility has since been
used successfully on other panels; therefore, although it may appear to be
an arbitrary assessment, the results justify its use until a better method of
calculation can be found.
Panels that have been cradled have frequently been thinned or have had
some surface preparation to enable the cradle to be tted. While such pre-
vious changes may have contributed to harmful eects suered by the
panel, they also make the attachment of another auxiliary support rela-
tively straightforward.
Recently, conservation work was undertaken on a panel for which it
was appropriate to use a exible attached auxiliary support. The panel had
not, however, been cradled or thinned, and consequently, the attachment of
the support to an irregular surface presented some dicult problems.
Description
The seventeenth-century Flemish painting Death of Orpheus, by Alexander
Keirincx and Roelant Savery,
10
measures 1.4 2.03 m; it is made up of six
oak boards with doweled and glued horizontal joins. Early in its history,
following some poor board rejoins, an attempt was made to atten the
panel. Four rigid poplar battens, each 100 mm wide, were glued into
trenched rabbets across the grain of the boards. Shrinkage of the boards
had then caused partial disjoins and some fracturing. In a misconceived
attempt to prevent further damage, buttery cleats were inserted across
the board joins, while the cross-grain battens were left in place. These
cleats were deeply recessed, with their grain perpendicular to that of the
boards. As would be expected from these contradictory interventions, fur-
ther damage had occurred in the form of fractures at the outer edges of
the butteries.
Some of the small buttery cleats had been removed and even
larger ones inserted, causing further fracturing. When the glued surface
joins of the battens failed, the battens were reglued and their ends screwed
to the outer edges of the panel. In one area on the bottom board, this had
recently caused a severe fracture 35 cm long (Figs. 712).
At various times during these conservation attempts, areas of the
boards had been crudely thinned, particularly where the large butteries
were inserted. Otherwise, the boards retained their original thickness,
varying between 6 mm and 10 mm, with consequent steps of up to 4 mm
at the joins. When the panel arrived for treatment, it showed signs of being
highly stressed. When viewed from the front, it was concave, and some
fractures were held open, indicating severe tension.
Before any structural work could be carried out, the panel was
rst kept in an environmental enclosure at 75% RH. When equilibrated, its
prole indicated that much of the high stress was relieved. The battens,
along with twenty-eight small buttery cleats and ve large ones, were
then removed so that rejoins could be made. The recesses from which the
Second Case Study:
Support for a Panel with
an Irregular Surface
392 Marchant
393 Tnt Dtvtiorxtr or a Fitxi tit Arracntn Auxi ii ar Surrorr
Fi gure 7
Alexander Keirincx and Roelant Savery, Death
of Orpheus, seventeenth century. Oil on oak
panel, 1.4 2.03 m. Private collection,
Northumberland. View before cleaning and
restoration, showing disjoins and fractures.
Fi gure 8
Keirincx-Savery, Death of Orpheus. The reverse
before panel work.
Fi gure 9
Keirincx-Savery, Death of Orpheus. This detail
before cleaning and restoration shows a
recent fracture in the bottom board.
Fi gure 10
Keirincx-Savery, Death of Orpheus. Detail of
the reverse before panel work, showing the
end of a batten that had been reglued to the
panel, a procedure that caused the fracture
shown in Figure 9.
cleats were removed were subsequently lled with shaped oak sections
with their grain in the same direction as that of the panel. Other buttery
cleats that did not require removal were planed down ush with the
panels surface.
Diculties of attaching a support to an irregular surface
After completion of all necessary structural repairs, the panel still presented
a formidable combination of problems. There were many faults and lines
of weakness. The panel was large and heavy, weighing more than 30 kg,
but in some places it was very thin and its surface totally irregular. It was
essential to provide reinforcement and to restrain the rapid response to
variations in RH by warping, to which the panel was now prone (Fig. 13).
To function properly, the secondary support would have to be in close
contact with the panel surface.
One of the fundamental principles of the support design is that
the calculated exibility of the battens should not vary from one to
394 Marchant
Fi gure 11
Keirincx-Savery, Death of Orpheus. Detail before
cleaning and restoration, showing a board dis-
join, with two lines of fractures below caused
by small and large buttery cleats.
Fi gure 12
Keirincx-Savery, Death of Orpheus. Detail of the
reverse before panel work, showing the cleats
that caused the fractures shown in Figure 11.
Fi gure 13
Keirincx-Savery, Death of Orpheus. The reverse
after structural conservation.
another. This could not be achieved if the battens were individually shaped
to the surface irregularities of the panel, a process that would create areas
of rigidity and weakness in the battens. Initially, therefore, they were made
identicalof uniform thickness and with a exibility calculated to provide
reinforcement. Calculations were made on the basis of using ten exible
battens, and it was decided to use one retaining strip on each of the six
boards. Sitka spruce was again chosen as the most suitable timber from
which to make the lattice.
With the layout for the main elements of the lattice decided, the
panel was then laid facedown on a horizontal surface with support to
maintain its camber established at 55% RH. The prepared battens were
laid across it at the chosen spacing and weighted to deect into contact
with the concave back surface of the panel. With the top surface of the
highest batten as a datum, the others were raised to the same level using
suitable packers.
When all of the battens conformed to a uniform curved plane,
the retaining strips were laid across the battens at the designated spacing.
The retaining blocks, which had been prepared oversized (in terms of
height), with slots already cut, were reduced in height and their bases
shaped to suit the position in which they would be glued to the panel,
with the slots aligned to engage on the retaining strips. This was a tedious
process involving 132 blocks, but it was important that it be done accu-
rately so as to ensure that the retaining strips would slide freely into place.
The packers supporting the battens were removed and replaced
with a balsa thicknessing layer glued cross-grain to the underside of each
batten. This layer was shaped to the surface prole of the panel. When
completed, the addition of the balsa was found to have no measurable
eect on the comparative exibility of the battens. The battens were now
all engaged by the retaining strips with a reasonably consistent contact
over the irregularities of the panel surface.
To complete the support, the battens were linked together with
two supporting strips to form a lattice, and an angle section of timber was
produced to act as a support for the weak bottom edge of the panel. This
angle was glued and doweled to the tips of the lattice, with bamboo pins
cut from swab sticks as dowels.
Framing and retention
Now that there was an even surface alignment of the battens, the produc-
tion and mounting of back springs was quite straightforward. The springs
consisted of exible bridging strips mounted centrally on each batten with
Plastazote-foam-cored timber pads. The space available gave the springs a
span of more than one-quarter of the batten length.
11
The use of pressure
pads was unnecessary, as it was proposed to use a retaining bar that could
bear directly against the bridging strips (Fig. 14).
With the auxiliary support engaged, the panels restrained warp
was monitored until stabilized at 55% RH, and the edge proles of the
panel were then recorded. A slip addition for the frame rabbet was made
to follow the panels proles. Alterations were also made at the back of
the frame to build up the rabbet. These alterations provided greater depth
to accommodate possible increased curvature in the panel and support
assembly of up to 30 mm.
395 Tnt Dtvtiorxtr or a Fitxi tit Arracntn Auxi ii ar Surrorr
A rigid timber beam, 100 30 mm in section, was then used
across the back of the frame as a retaining bar to hold the panel/support
assembly in place (Fig. 15).
Finally, the back of the frame was totally enclosed with two thin
plywood sheets as backboard sections, tted above and below the retaining
bar. The backboards may be removed to allow inspection of the retained
assembly without its being disturbed in any way.
An advantage of using this type of auxiliary support system is that
it is one of the least intrusive methods of tackling problems such as those
presented by the Keirincx-Savery panel. Most of the remaining original
surface features have been preserved, and if at any time there is a suspi-
cion that further problems are arising, conservators can gain access
quickly and easily by removing the lattice, leaving only the retaining
blocks attached to the panel. By themselves, these blocks are unlikely to
have an adverse eect on the panel and do not preclude the possibility of
further conservation work being carried out, after which the lattice could
again be easily replaced.
Reducing friction on the supporting edge of heavy panels
When the panel work was completed, there still remained a framing
diculty to overcome. The Keirincx-Savery highlighted this recurrent
problem of displaying large, heavy, horizontal-grain panels.
396 Marchant
Fi gure 14
Keirincx-Savery, Death of Orpheus, reverse. The
retaining blocks are glued in place; the exible
auxiliary support is engaged.
Fi gure 15
Keirincx-Savery, Death of Orpheus, reverse.
The completed panel/support assembly is
shown mounted in the frame, with the retain-
ing bar in place.
Even with the achievement of a exible auxiliary support that will
allow changes of curvature (although partially restrained) to occur in a
panel, the whole object of the exercise will be defeated if the panels sup-
porting edge gets stuck and cannot move smoothly in the frame rabbet.
With lightweight panels it has been common practice to use Teon/PTFE
pressure-sensitive adhesive tape to line the tray or frame rabbet, thus
reducing friction against the load-bearing edge of the panel. With large,
heavy panels, the reduction in frictional resistance achieved by Teon tape
may only be sucient to prevent total jamming. Movement of the panels
bottom edge is still likely to be erratic, however, with sudden jumps occur-
ring only when the warping stresses build up in the panel and exceed the
frictional resistance imposed by its weight. Also, it is not uncommon to
nd environmentally responsive panels that have warped away from a slip
prole and have become wedged at the back of the frame rabbet.
A solution to this problem of reducing friction, found suitable for
the Keirincx-Savery panel, was simply to mount the bottom supporting
edge of the lattice on bearings. Several bearing designs were investigated.
Among them, linear slide bearings were found to be available with
coecients of friction as low as 0.003 (i.e., a force of 3 gm will move a
1 kg load on a horizontal surface). These bearings are high-specication
devices for engineering applications and as a result are relatively expensive.
For the Keirincx-Savery panel painting, however, the type chosen
were simple bearings known as Ball units that were found to work
extremely well and are being considered for use on some even larger pan-
els. A possible disadvantage of Ball units is that the minimum dimension
below the panel needed to accommodate them is 20 mm, whereas with
linear slide bearings it can be as little as 8 mm. Fortunately, the Keirincx-
Savery frame was substantial enough for 20 mm deep recesses to be cut
for the bearing to run in. Two Ball units were used, giving a combined
specied load-bearing capacity of 50 kg. Polished 18-gauge stainless steel
blanks were placed in the recesses as a running surface for the bearings.
If adequate depth had not been available in the frame, then the thinner,
more expensive type of linear slide bearing would have been considered
(Figs. 16, 17).
Since completion of the restorations,
12
the Keirincx-Savery
panel/support assembly, mounted in its frame, has been monitored at the
397 Tnt Dtvtiorxtr or a Fitxi tit Arracntn Auxi ii ar Surrorr
Fi gure 16
Detail of the vertical support bearing.
Fi gure 17
The support bearing seen from below.
authors studio. So far, the results of the structural conservation work look
very promising. The eorts made to ensure the long-term stability of this
panel painting will have likely been worthwhile (Fig. 18a, b).
When a secondary support is attached to a weak, responsive panel, it
fullls two functions. One is reinforcement, the other is restraint. Restraint
is the function that is potentially damaging and also the most dicult to
evaluate. It may be dened as the degree of rigidity required to resist the
bending force of the panel. If the resistance is too high, the panel may
be damaged.
A safe level of resistance could be calculated with basic engineer-
ing formulas if the panels bending force can be found, but this calculation
requires a gure for the modulus of elasticity (E value) across the grain of
the panel. Approximate E values perpendicular to the grain may be derived
from reference tables, but only for sound timber samples. For aged, stress-
weakened, or damaged timber, these gures are not relevant and cannot be
used. If the panels strength cannot be estimated, then it is virtually impos-
sible to calculate the rigidity of battens needed for tolerable restraint.
An alternative approach is to consider the problem from the point
of view of reinforcement. This assessment can be made with the panel
lying horizontally over a central beam, with the battens providing the
rigidity necessary to support the panels weight without it deecting too
far. Calculating reinforcement in this way is relatively easy, and the judg-
ments involved are not too demanding.
In practice, it has been found that a support with a exibility cal-
culated for reinforcement also provides the safe level of restrainta level
that was dicult to determine by other methods.
If battens, which have been made up to the calculated dimensions
with a uniform section, are now tapered in width from the center to the
ends, their rigidity will decrease progressively away from the center. The
bending force that the panel exerts on the battens also reduces progres-
sively from the center to the outer edges. Therefore, the resistance to
bending imposed by the battens on the panel will be balanced, producing
an even restraint across the width of the panel. As a result, when the sup-
port battens are attached to the panel, the tension on all of the retaining
blocks will be more equally distributed than if the battens were left as a
uniform section.
The deection calculated for point loading at the end of the bat-
ten will increase by about 50% after the batten is tapered. This increase
does not constitute an error in the method of calculation, as it is compen-
sated for by the actual load imposed by the panels weight being uniformly
distributed, so that a corresponding reduction in deection is produced.
With this method of calculation for batten dimension, the support
system has been applied to several panels that varied considerably in size,
weight, thickness, and timber type. In all cases, the measurable reduction
in curvature after the supports were engaged has been 30% or less. This
level of restraint is judged to be below the threshold where damage is
likely to be caused.
The support also provides a degree of reinforcement, enabling
the panel to support its own weight and to be handled safely and with
more condence.
Summary of the Principles
of Calculating Batten
Flexibility
398 Marchant
In general, before making a commitment to a detailed design, the panel
conservator must amass all available information. It should be possible to
specify the dimensional limits of movement of the panel that will deter-
mine tray depth, and so on; this can be done by monitoring movement.
More information may be gained from assessing previous damage to the
panel and painted surface, as well as from assessing conditions under
Conclusion
399 Tnt Dtvtiorxtr or a Fitxi tit Arracntn Auxi ii ar Surrorr
Fi gure 18a, b
Keirincx-Savery, Death of Orpheus. The general
view (a) and a detail (b) show the paintings
condition after restoration.
a
b
which the panel may be kept in the future. Problems could also develop,
especially in the ground and paint layers, when unrestricted freedom of
response to environmental changes is allowed.
Sometimes the solution to the problems may be a compromise
dictated by display requirements. There is little point in designing a micro-
climate box or a 15 cm deep tray that cannot be accommodated in an orig-
inal frame or is unacceptable to the client for display purposes.
It is also worthwhile to consider a combination of ideas rather
than a single solution. For example, it should be possible either to reduce
or to slow down the response of a panel to environmental conditions with
a choice of barrier or buering techniques, and then to combine the cho-
sen technique with a restraint or an auxiliary support. In addition, there is
now a wide availability of technology that makes environmental control
possible and more cost-eective in buildings where it would not have been
considered previously.
It is not easy to generalize or adopt a standard practice when
deciding which method to use. Every panel is dierent, and it would be
incorrect to expect that an acceptable answer to one particular problem
can be adopted as a principle for general use.
The fashionable answer among some nineteenth-century cradle
makers was to thin, atten, and restrain panel paintings so that they could
be displayed like canvases. Today our views are dierent, and a lot of time
is spent removing work that, when executed, was thought to follow the
correct approach but that can now be seen to be damaging. To avoid
falling into the same trap, todays conservators should adopt an open-
minded approach and continually reappraise their methods and learn from
their own experience and that of others.
It is the authors belief that many conservators might remain isolated from
the benets of an exchange of ideas if the opportunity to meet other spe-
cialist conservators were not made available. It is greatly appreciated that
institutions such as the Getty Conservation Institute continue to provide
these opportunities at an international level. The author would also like to
express his appreciation to the British Standards Institute (BSI) for his use
of material from a BSI publication.
1 Reference tables of the modulus of elasticity for timbers including Sitka spruce appear in
Molesworth 1951:43235.
2 Sitka spruce (Picea sitchensis), a softwood imported from Alaska and Russia, having consistent,
reliable mechanical properties. It is used for structural framework in some light-aircraft con-
struction (see Keen 1919).
3 Reference tables for modulus of elasticity from Molesworth (see n. 1) are given in lb in
2
; they
have been converted into n mm
2
by multiplying by 0.0068947.
4 PVA Evo-Stik wood adhesive is generally the preferred choice for structural work. It is consid-
ered to have good long-term stability and exibility, giving it higher shock resistance than ani-
mal glues, which may become brittle with age. Other adhesives used in these case studies were
rabbit-skin glue, for replacement of buttery cleats, and an impact adhesive containing
toluene, for bonding Plastazote polyethylene foam to timber.
Notes
Acknowledgments
400 Marchant
5 If board width is sucient, it would be preferable to use three rows of slotted retaining blocks
per board. This provides the best pattern of restraint against warp of each individual board.
With narrow boards where space is sucient for only one row of blocks, it is better to place
them near the center line to avoid creating tension close to the board joins. It is not consid-
ered advisable to use the blocks to reinforce or span board joins, a practice that can frequently
be seen with xed-cradle members.
6 Teon/PTFE skived tape, with a pressure-sensitive adhesive coating on one side, has been
found to be the best of the range of PTFE products for reducing friction. A cheaper alterna-
tive recently found available is polyolen tape. This is an ultrahigh molecular weight (UHMW)
polyethylene material with a coecient of friction comparable to PTFE. The pressure-sensitive
rubber adhesive coating is more suitable for timber, and it also has improved mechanical char-
acteristics, such as lower elongation and higher wear resistance to abrasion. As yet, it has not
been in use in the authors studio long enough for full evaluation.
7 In the United Kingdom, this means that the panel has been stabilized at 55% RH.
8 Plastazote is a closed-cell, cross-linked polyethylene foam available in a number of densities.
The one used as a core in the timber mounting pads is the low-density LD24.
9 Tables of values of modulus of elasticity of timber relate to data obtained from testing in a
direction parallel to the timber grain. Figures do not exist for E values perpendicular to the
grain. However, a useful reference can be found in a British Standards Institute (BSI) publica-
tion (1991:pt. 2, clause 11 [Additional Properties]): In the absence of specic test data, it is
recommended that, for tension perpendicular to the grain, torsional shear and rolling shear,
values which are one-third of those parallel to the grain should be used. For modulus of elastic-
ity perpendicular to the grain, a value of one-twentieth (i.e., 0.05) of the permissible modulus of elas-
ticity should be used (emphasis added).
Properties of Sitka spruce are given in table 11 of the BSI publication. Information for
obtaining complete copies of the publication can be found in the Materials and Suppliers
section below.
10 Alexander Keirincx (16001652) and Roelant Savery (1576?1639). The painting depicts
Orpheus, who could enchant the beasts, being attacked by the Thracian women.
11 A span of not less than one-quarter or more than one-third of the batten length has been
found in practice to be a good dimension for which to aim.
12 Restoration of the painting was carried out at Lank Sanden Studio in London.
Ball units, Alwayse Engineering Ltd., Warner Street, Birmingham B12 0JG, England. (Large
range of Ball transfer units available; the type in use are from the Solid Body Unit range.)
BSI Publications, BSI Customer Services, 389 Chiswick High Road, London W4 4AL, England.
Evo-Stik, wood adhesive, waterproof, or extra fast resin W, Evode Ltd., Common Road,
Staord, England.
Linear-motion slide bearings, SKF Engineering Products Ltd., 2 Tanners Drive, Blakelands,
Milton Keynes, MK14 5BN. (Small units are available in the standard slide range, RM series.)
Plastazote (a closed-cell, cross-linked polyethylene foam, REF LD 24), BXL Plastics Ltd., Mitcham
Road, Croydon, Surrey CR9 3AL, England. (Distributed by Hemisphere Rubber Co., 65 Fairview
Road, Norbury, London SW16 5PX, England.)
Polyolen ultrahigh molecular weight (UHMW) polyethylene tape with a pressure-sensitive
rubber adhesive (marketed as Polycohr), CHR Industries, Inc., 407 East Street, New Haven,
CT 06509. (The European supplier is Furon CHR Products, P.O. Box 124, 7640 AC Wierden,
Netherlands. Distributed in the United Kingdom by Polypenco Ltd., now part of DSM
Engineering Plastic Products UK Ltd., 83 Bridge Road East, Welwyn Garden City, Hertfordshire
AL7 1LA, England.)
Teon/PTFE (polytetrauoroethylene) tape with a pressure-sensitive silicon adhesive
(marketed as Temp-r-tape HM series), CHR Industries (see information for polyolen tape).
Materials and Suppliers
401 Tnt Dtvtiorxtr or a Fitxi tit Arracntn Auxi ii ar Surrorr
British Standards Institute
1991 Publication no. BS5268. London: BSI.
Brough, J., and J. Dunkerton
1984 The construction of panel trays for two paintings by the Master of Cappenberg.
National Gallery Technical Bulletin 8:6370.
Keen, G. R.
1919 Aeroplane Timbers: Their Structural Formation and Mechanical and Commercial Properties.
London: Rider and Sons.
Molesworth, G. L.
1951 Molesworths Handbook of Engineering Formulae and Data. 34th ed. London:
E. and F. N. Spon.
Thomson, G.
1978 The Museum Environment. London: Butterworths.
References
402 Marchant
T
nt ari oai oaiitr, iono, has a comprehensive collection
of western European paintings from the thirteenth to the twentieth
century. There are some one thousand panels in the collection,
more than half of which are Italian and painted on poplar. The other main
schoolsDutch, Flemish, and Germanusually used oak. Other woods
used include lime and beech (used by Lucas Cranach the Elder, for example),
walnut or fruitwood (pear), and pine.
The National Gallery has mostly conventional panel structures of
dierent types of wood with members glued together, mostly with animal
glues, and usually with the grain running in the same direction as the
joins. There are also some complex structures, of which Rubenss panels,
such as A View of Het Steen (NG66) and The Watering Place (NG 4815), are
prime examples (Brown, Reeve, and Wyld 1982) (Fig. 1).
Most of these panels have undergone some form of conservation
work, ranging from crack repair to added buttons, battens, and cradles, or
thinning and transfers. For the most part this work has been carried out
prior to or at the time of acquisition by restorers abroad and in England.
403
Anthony M. Reeve
Structural Conservation of Panel Paintings at the
National Gallery, London
Fi gure 1
X ray of Peter Paul Rubens, The Watering
Place, 161522. Oil (identied) on panel,
99.4 135 cm. National Gallery, London
(NG 4815). The eleven panel members, as
well as the buttons and battens, are seen
before conservation treatment.
The National Gallerys Conservation Department, founded in
1946, initially occupied two converted exhibition rooms. Restoration
studios built specically for this purpose were opened in 1959. In theory,
in-house restorers have carried out all the work on the collection since
1946. In practice, the records show that during the early years, there
was still considerable structural work carried out by private restorers
(Morrill is the most often mentioned). In 1949 (as described below in
connection with balsa-wood buildup), Richard Buck came to the gallery
from the United States with new ideas on panel work and transfer. In
1965 the gallery was still inclined to the removal of original wood,
believing it would minimize the possibility of further movement; com-
plete transfer was sometimes considered. Treatments of various kinds
have been developed over the years, progressing to the present day. In
looking back, one can see that some of the conservation treatments
may not have been the most eective, although they were accepted prac-
tice at the time. The author has supervised all the structural treatments
in the department since 1977.
This article contains a description of the methods used in the
National Gallery at present. Where relevant, old methods and materials
are discussed. As a general rule, every part of the original support is pre-
served whenever possible. Necessary treatments are designed to be as
easily reversible as possible. Old methods and materials of conservation
are not changed unless new ones can be shown to be more satisfactory.
The best environment for panels is considered to be 55% relative
humidity (RH) at 21 C; it is preferable to err on the side of higher, rather
than lower, humidity. It is best never to move panels from these conditions
if possible. The transport of panels from one country to another by air-
craft and the exposure to a dierent, usually drier, environment have been
prime causes of much panel movement and subsequent deterioration.
Deterioration is even more pronounced if restrictive conservation has
been carried out rst. The location (e.g., church, country house, museum)
of a panel greatly inuences the types of treatment and materials neces-
sary to carry out the best conservation.
Animal infestation
Any suspicion of worm or beetle activity should be treated to prepare the
individual object for conservation, as well as to protect other objects from
infestation. Various forms of treatment (gassing, oxygen deprivation, or
liquid application) are suitable for particular problems.
Surface consolidation
Sturgeon glue, normally diluted to an approximately 5% solution, is com-
monly used with controlled-heat spatulas for conserving loose or blistered
areas. If this proves unsuccessful, one may have to use a dierent adhesive
to secure old aking or impregnations. After surface consolidation of a
painting that has previously been restored, it is usually preferable, where
possible, to clean the painting to remove excessive llings that might
impede structural consolidation. Surfaces can often be improved where
an old conservation treatment was not totally satisfactory.
Present Conservation
Methods
404 Re e ve
Facing of the surface before structural consolidation work
Panels once consolidated on the surface are usually faced before any other
treatment is carried out. Crack or join repairs are usually faced up to their
edges. The facing should cover the surface entirely if structural or removal
work is to be carried out on the back. The rationale for choosing a particu-
lar facing material and facing mixture depends on the surface, solubility,
and condition of the painted layer and also on the structural work to be
carried out. The materials commonly used include Eltoline tissue with
Paraloid B72 or B67, or damar with a little wax. Occasionally, aqueous fac-
ing adhesives are used, but usually only for transfer treatment. If more
than one facing has been applied and it is necessary to release or remove
one or more of the facings, then the later layers should have dierent
adhesives to ensure that the picture will always be protected. Where there
are open cracks to treat, and protection is necessary, B72 or B67 is nor-
mally used rst.
Removal of old nails or xings and the treatment of
cracks and joins
If normal methods of removing old nails or xings are not adequate, heating
the metal (which causes expansion and the ensuing contraction) may help.
Having used traditional clamping tools and experienced their limi-
tations, the author designed a clamping table, which was manufactured by
Willards of Chichester (Reeve 1990) (Fig. 2).
1
The adhesive generally chosen for joining cracks is Cascamite, a
powdered urea-diformaldehyde synthetic resin with a hardener. Its advan-
tages are that it produces bonds that perform well when exposed to
extreme dryness or dampness, or even when completely saturated in
water. The aqueous quality of Cascamite allows softening and slight
expansion of the edges of wood being joined. It also has the possibility of
being used in dilute form for penetrating small closed cracks, or in thicker
405 Srrucrurai Cos trvari o or Pati Pai ri os ar rnt Nari oai Gaiitr, Lono
Fi gure 2
Clamping table for panel conservation.
concentration for open joints and wider cracks. If the cracks are over a few
millimeters wide, rye our can be added as a ller; if necessary, polyvinyl
acetate (PVA) dispersions can also be used to reduce the brittleness. A wet-
ting agent such as Oxgall can also improve adhesion when permeation is
not sucient. Cascamite has a two-hour or longer working time, making
it very useful for working with the nal alignment of deformations.
Cascamite is quite a brittle adhesive, although it is adequate to
cope with the natural movements of a panel if joined and used properly.
Subsequent applications are possible in inaccessible areas, so should
diculties arise, it will rebond very well. Also, changes in RH should not
produce the same magnitude of dimensional change that proteinaceous
glues undergo, and it is not susceptible to attack by microorganisms.
Where possible, Cascamite is applied to both sides of the join. If
the join is partially sealed or only slightly open, the adhesive is applied
along the join back and front alternately, and the panel is exed sideways
or up and down as much as the structure will allow without causing fur-
ther cracking. This action creates an absorption of the adhesive and expul-
sion of the air. A dabbing movement on the surface can also be eective.
In some cases it may be necessary to use another adhesive. PVA
dispersion emulsion (Resin W) is occasionally used; however, it is less easy
to work with than Cascamite, as it has a rather short drying time of ten to
fteen minutes. The National Gallerys Scientic Department frequently
reviews new materials in search of alternatives. For larger cracks, wood
(preferably of similar age and type) may be inserted, with the grain run-
ning with the original.
A variant of a widely practiced method used initially for trying
to correct warping and then for reinforcing cracks was used at the gallery
for a while in the late 1950s and early 1960s. A V-shaped router was made
and set to the desired depth to cut a groove along the line of the crack at
the back of the panel, removing the original wood. Another tool was then
used to produce a V-shaped wedge to t into the newly cut channel,
either in long straight strips or in short strips if the cracks were irregular.
The idea usually behind this was to penetrate through to the back of the
ground and to produce two new side surfaces to bond to the V-shaped
wedge; this method is no longer used, however. In accordance with the
ideal of preserving as much of the original wood as possible, cracks are
joined edge to edge whenever feasible.
Cases involving insect attack or dry rot may require the removal
of the original wood to consolidate the panel; however, this procedure has
rarely been necessary on artworks in the National Gallery collection.
Moisture treatments
After crack consolidation or release from previous restrictions (for example,
removal of battens or a cradle), a panel may adopt a greater concave or
convex warp. It may be possible to reduce the warp by exposure to mois-
ture and relaxation under varying pressure over a period of days or weeks.
The low-pressure conservation table, using circulated moisture under a
controlled vacuum, is becoming an alternative for this treatment (Reeve
1984; Reeve, Ackroyd, and Wright 1988) (Fig. 3). This table and its use are
described in more detail below, in the account of the panel treatment for
Cosm Turas Annunciation.
406 Re e ve
Fi gure 3
Multipurpose low-pressure conservation
table with small warped test panel during
moisture treatment.
Consolidation and impregnation of
woodworm-aected areas
Where there are cases of woodworm attack, it is very dicult to consoli-
date the remaining wood, especially immediately behind the paint. The
worst cases of this may eventually lead to the necessity for a transfer.
Various materials have been tried in impregnation tests and evaluated for
their ecacy in penetration and consolidation, with Paraloid B67 in white
spirit found to be the most suitable. This material also could and would
act as a moisture barrier, in preference to the old methods of applying
Saran or hot wax. B67 and wood our are used for inlls of any large open
wormholes or lost areas. Very large losses would possibly be inlled with
wood similar to the original, with the grain running in the same direction
as that of the original.
Moisture barriers
To create a moisture barrier by means other than impregnation with
Paraloid B67 (for example), a layer of material preimpregnated with Beva
371 could be attached to the back of the panel with a warm spatula. This
technique can also give extra support to the panel, reducing the need for
further treatment.
Inlls of balsa wood
Where it proves necessary to remove restricting bars, battens, buttery
buttons, cradles, and so forth from the back of the panel, it is customary
to inll with a material such as balsa wood (Fig. 4), cut to half its depth
across the grain at 2.5 cm intervals to counter any tendency of its own
to move, and usually running parallel with the grain of the original.
Sometimes original chamfered sliding battens can be reduced a little, also
cut halfway through at 2.5 cm intervals, and reused.
Panel trays
Where the original panel is in a state too fragile to support itself, either
because of thinning or because of inherent weakness, it is often incorpo-
rated into a tray. The tray is a secondary support that has been used in the
National Gallery for a long time, although its construction and materials
have been improved and developed in recent years.
The panel tray consists of a backboard made up of Aerolam F
board (aluminum honeycomb covered in a resinated berglass) with the
internal edges cut back to allow the inset of a cedar strip (Brough and
Dunkerton 1984; Dunkerton and Smith 1986), the purpose of which is to
attach the panel tray to the outer oak frame, which is made to cap the
front edges of the picture (Fig. 5). In a tray, the picture is completely sup-
ported at the back on blocks (minimum 6 mm thickness) of either Evazote
(low-density polyethylene [LDPE] copolymer foam) or Plastazote (LDPE
foam), with at least 3 mm of the same material under the oak strip that
caps the sides and the edges. Evazote and Plastazote are available in
dierent densities, and the strip of Evazote or Plastazote at the bottom
of the tray frame supporting the picture should be of a higher density to
prevent it from slipping down in the trays rabbet.
407 Srrucrurai Cos trvari o or Pati Pai ri os ar rnt Nari oai Gaiitr, Lono
Fi gure 4
Zanobi di Benedetto Strozzi, The
Annunciation, ca. 1450. Reverse. Tempera on
panel, cut on all sides, 103.5 141.6 cm.
National Gallery (NG 1406), London. On the
back of the panel, inlls of toned balsa are
seen on the right side; balsa pieces ready for
tting are on the middle and the left side.
The Evazote/Plastazote is shaped to accommodate potential
panel warp. Only minimal rows of the foam blocks are used, to allow
exing of the panel during environmental changes. This, of course, may
happen not only at the edges but anywhere across the width or length of
the panel, depending on its structure: restriction of movement is kept to a
minimum by this means.
The tray acts as a very substantial protection to a fragile panel,
both in its frame and during handling. If the environments are expected to
vary, slots can be cut in the tray backboard to allow greater freedom of air
movement and to reduce the possibility of concave warp from the front.
These slots should be covered with a porous material such as polyester
net. However, it is essential to have enough movement available for the
panel in the tray through use of blocks and edge slips that are suciently
exible. These trays can usually be accommodated in the original frames
with a little adjustment, and the front edge of the tray can be toned or
gilded to form the inner rabbet of the frame.
Balsa-wood buildup
Balsa-wood buildup is often necessary when, following the removal of a
cradle or other veneered additions, a panel is too thin or weak for a tray.
The most commonly cradled panels are on poplar and are often thinned to
less than a third of their original thickness.
After a panel is released from a cradle and the cracks are consoli-
dated, it usually adopts a convex warp when seen from the front and may
also be too thin or too big to maintain a at or near-at conformation.
After moisture treatment where required, it may prove necessary to attach
408 Re e ve
Polyethylene shaped to
edge and face of panel
to accommodate movement
and warping
Cedarwood insert
to provide fixing
Oak tray frame
Fixing screws
Polyethylene
buttons
cut to support
the panel
Panel picture face
Ventilation slot in backboard
showing aluminum honeycomb core
Slot is covered by a porous material
Higher-density polyethylene
at base of picture
Gap to prevent
distorting edge packing
when placing backboard
Fiberglass and epoxy resin
facing layers to board
Fi gure 5
Construction elements of a panel tray.
a secondary support to the back, which will normally return the panel to
its original thickness or even make it slightly thicker.
This procedure used to involve an updated and improved form of
a methodthe balsa-wood and wax-resin cement buildupintroduced
from the United States by Richard Buck in 1949. This method has been
described in the National Gallery Technical Bulletin (Smith, Reeve, and
Ashok 1981) (Fig. 6). Since then, the method has been improved by the use
of a dierent materials as an interleafimpregnated with Beva 371 on
both sidesbetween the original panel and the buildup, thereby prevent-
ing impregnation of the wax-resin into the original panel. Also, the balsa
planks are all sawed halfway through at 2.5 cm intervals after the applica-
tion of each layer, in order to reduce their strength (Fig. 7).
The application of the modied version of a balsa-wood panel
buildup begins after moisture treatment or attening, where necessary.
New renements of the method using the multipurpose low-pressure table
are described in the case study below.
Transfers
Transferring a painting is the last resort and is considered only when the
support or ground is no longer able to maintain the painting. Methods
vary according to the problem. The only example carried out at the
National Gallery in recent years was the transfer of The Incredulity of Saint
Thomas (NG 816) by Cima da Conegliano (14591517), in which the fol-
lowing procedure was employed (Wyld and Dunkerton 1985).
After removal of the remaining wood and consolidation of the
ground from the back, a reversible isolating layer of acrylic primer was
applied, followed by a vinyl emulsion ller. An interleaf of nely woven
white linen stretched on a loom was coated on both sides with a synthetic,
heat-bonded adhesive (Beva 371) and attached to the reverse of the paint
and ground. This was, in turn, attached to an aluminum honeycomb
epoxy-coated berglass board (Aerolam F board), also coated with Beva
371. The author has found it more aesthetically pleasing to use a slightly
textured surface for these supports; a at texture seems to impose an
unnatural smoothness.
Panel ttings
For support, early panels or fragments may need specially designed brack-
ets of metal or other material, lined with polyethylene foam or velvet, so
that no xings are applied into the original panel. The security of the
object must also be a consideration in the design of the brackets.
Frame tting and exhibiting
A picture should be put into the frame against a soft surface of velvet or
similar material to prevent scung of the edges. Panels that are warped
need shaped polyethylene foam strips between them and the rabbet. In
order for the foam strips not to become compressed at the base of the
panels, they must be made of a higher density polyethylene or of balsa
wood. Panels should be held in frames with as few ttings as possible, with
adequate exible polyethylene pads between the ttings and the panel.
The ttings should also be placed at the ends of the wood grain only
at the top and bottom for vertical grains and at the sides for horizontal
409 Srrucrurai Cos trvari o or Pati Pai ri os ar rnt Nari oai Gaiitr, Lono
grainsand toward the center of the panel. The back of the frame should
always project beyond the picture to prevent the panel from pressing
directly against the wall. Also, a backboard of some sort helps to act as
an environmental buer and to prevent accidental damage.
When the panels are housed in an uncontrolled or uctuating
environment, it may be necessary to incorporate the panel and/or panel
and framewhether in a tray or notinto a vitrine (to assist in reducing
the uctuation of temperature and RH between the panel and surround-
ing air) or into a climate-controlled exhibition case.
This small (45 34 cm) panel is a fragment of the Annunciation by Cosm
Tura (143195), probably painted around 1480. The picture is on a poplar
panel painted up to the edges and clearly cut all around. It was acquired by
the National Gallery in 1874 and recorded to be in good condition. In 1915
the old parquet (cradle) was removed, and the breaks were reset (through
the head in a vertical line and elsewhere). The panel was then veneered,
and a new parquet was applied. In 1991 the picture was proposed for clean-
ing and restoration, procedures that were carried out by Jill Dunkerton. The
structural work was done by the author and David Thomas.
2
Case Study: Treatment of a
Painting by Cosm Tura
410 Re e ve
8. Canvas ironed onto the back and over the sides
of the balsa-wood buildup with wax and resin
7. Second layer of balsa wood laid
at right angles to the first layer and
scored on the underside as a key
5. First layer of balsa-wood
planks laid with the grain
running in the same direction
as the original panel and
scored on both sides as a key
6. Wax, dammar,
resin, and wood flour
poured over the first
layer of balsa wood
4. Wax, dammar, resin,
and wood flour poured
over the hessian interleaf
The balsa-wood planks
are agitated into position
to disperse the air and
level out the wax mixture
1. Original panel
3. Open-weave hessian
(burlap) ironed into wax
and resin on the back
of the panel
2. Wax and dammar resin
ironed into the panel
Fi gure 6
Diagram of the old method of balsa-
wood buildup.
Photographic examination by infrared and X ray was carried out
to estimate the true condition of the remaining panel and paint (Fig. 8).
Infrared photography showed that there was extensive restoration down
the o-center vertical crack or join that runs vertically through the Virgins
face, as well as on some other, smaller areas of damage. The X ray showed
a very worm-eaten panel, in which most worm channels seemed to have
been lled with chalk, glue, and pigment. There were also several insets
of a dierent wood in the complex vertical crack at the top and bottom
edges. The original panel had been planed down to a thickness of no
more than 2 mm. It was surrounded by thin oak strips, veneered onto
mahogany, and cradled with oak sliding bars and mahogany xed battens.
The cradle had caused a slight concave warp on the length of the panel.
The paintings poor condition had been exacerbated by these past
treatments, which were causing further cracking, blistering, and aking.
The painting was also covered with a very discolored varnish. Restorations
covered original paint in some areas, and the surface was shown to be very
uneven under raking light. In order to improve these panel defects, extensive
panel treatment was proposed, involving the removal of all later additions.
After cleaning, the wooden inserts could clearly be seen from the
front (Fig. 9). Under raking light, it was also clear how badly the surface
had been aected, especially in the Virgins face. Before facing, a tracing
was made of all the major cracks and problem areas for future reference,
as well as to relate the work to the back of the panel.
411 Srrucrurai Cos trvari o or Pati Pai ri os ar rnt Nari oai Gaiitr, Lono
10. Canvas ironed onto the back of the
balsa-wood buildup with wax resin
9. Second layer of balsa wood
sawn across the grain
as the first layer
8. Second layer of balsa usually
laid at right angles to the first, scored
on the underside as a key
5. First layer of balsa-wood
planks laid with the
grain running in
same direction
as the original
panel and
scored on
both sides
as a key
6. First layer of
balsa-wood planks
sawn across the grain
to half their depth at 2.5cm
intervals after application
3. Stabiltex or polyester net
impregnated with Beva 371 and tack
bonded onto the interleaf (or panel)
2. Fine open-weave muslin impregnated
with Beva 371 and tack bonded onto the panel
Note: One or other interleaf (or both) can be used
7. Wax, damar,
resin, and wood flour
poured over the first
layer of balsa wood
4. Wax, damar,
resin, and wood flour
poured over the interleaf
The balsa-wood planks
are agitated into position
to disperse the air and
level out the wax mixture
1. Original panel
Fi gure 7
Diagram of the new method of balsa-
wood buildup.
First the picture was faced. With the goal of realigning uneven
fragments of the picture adjacent to the cracks, dierent resins were used.
The areas of paint 1.25 cm wide on either side of the main split were faced
with small pieces of Eltoline tissue and Paraloid B72 in xylene. The pieces
were shaped to support and protect the edges of paint along the split and
some islands of paint and ground within the split, while allowing the split
and other cracks to remain accessible. Two further complete layers of fac-
ing were applied over the whole surface with the Paraloid B67 in white
spirit. This facing protected the painting during cradle and veneer removal,
but when it was necessary to remove some parts during the crack conser-
vation, it could be done without disturbance to the B72 facings.
The mahogany cradle was removed by the procedures of sawing
across the glued battens at 2.5 cm intervals and chipping away with a
gouge or chisel.
This treatment exposed a mahogany veneer approximately 5 mm
thick, which was removed with hand gouges and scalpels. Once the
mahogany was removed, the back of the thinned panel could be seen
(Fig. 10). Many open cracks in the back of the panel had not been visible
from the front. The procedure also exposed the many worm channels, seen
in the X ray, that had been lled with pigment. The llings were removed
where necessary to enable realignment and securing of the cracks and old
joins. In some areas where the llings were removed, there was no panel
fabric left, and the back of the original gesso was exposed. It is not certain
when the picture was thinned: it could have been when the rst cradle
was applied, or possibly when the panel arrived in England. However, it is
thought more likely to have been during the second intervention; during
thinning, the panel collapsed in some of the worm-eaten areas, the inlls
412 Re e ve
Fi gure 8, above
X ray before treatment of Cosm Tura, The
Virgin. Fragment of the Annunciation, 147580.
Oil and egg (identied) on panel, 45.1 34
cm. National Gallery (NG 905), London.
Fi gure 9, ri ght
Cosm Tura, The Virgin. After cleaning and
before panel treatment, the painting shows
cracks, losses, and wooden inlls.
Fi gure 10
Cosm Tura, The Virgin, reverse. The bottom
right corner of the back of the panel shows
worm damage exposed after the cradle and
mahogany veneer were removed.
of wood were applied, and a second orange putty was pushed in from
behind around the new inserts next to the older white putty. These
were now strengthened with dilute PVA (Vinamul 3252) in dispersion.
Realignment of distorted parts was accomplished by softening and reopen-
ing some of the old joins and insets, gradually reweighting, drying, and
gluing them into new positions while the picture was placed facedown.
Voids and worm channels were lled with thin layers of Fine
Surface Polyllaa vinyl ester of Versatic 10 (Shell Resin)PVA copoly-
mer (Veo Va-PVA) with ller and thickener (Caley 1993). The mahogany
insets and oak strips around the edges were left in place as a protection,
but those along the top and bottom edges ran against the grain. Those
insets were sawed through at 1.25 cm intervals to prevent any restriction.
With all of the cracks glued and secured, the panel now took on a convex
warp when seen from the front.
Because the panel was exceptionally fragile, it was decided that a
balsa-wood buildup was necessary to provide support and stability. The
panel was treated with controlled moisture to reduce the warp that had
occurred after the removal of the additions and consolidation of the
cracks and joins. In a departure from the traditional method of suspend-
ing the panel over damp pads, treatment was carried out on the multi-
purpose, low-pressure conservation table, hitherto used primarily for
canvas treatments. The painting was placed faceup on the table and cov-
ered with Melinex (known in the United States by the trade name Mylar)
(Fig. 11). A very mild surface vacuum was applied, and the table was
warmed slightly to 30 C. Room RH was raised from 55% to 7580%.
The air circulated in the area under and around the panel; humidication
continued for about an hour.
The panel relaxed naturally, and as it did so, the surface vacuum
was increased accordingly. When the panel had relaxed completely,
humidication was turned o, the surface vacuum was maintained, and
the excess humidity was drawn away from below with the built-in
dehumidier, bringing RH back to 55% while slowly reducing the temper-
ature of the table to 21 C. The dehumidier was kept running at the
same setting for several hours. The vacuum was then turned o and the
panel left on the table until the next day, where it had attened consider-
ably, although it still had a slight frontal convex warp.
413 Srrucrurai Cos trvari o or Pati Pai ri os ar rnt Nari oai Gaiitr, Lono
Fi gure 11
Cosm Tura, The Virgin. Moisture introduc-
tion on the multipurpose low-pressure conser-
vation table.
Further moisture treatment from the back was necessary, but a
slower, more even drying process was desired. Therefore, moisture was
sprayed onto the back with a pressurized ne-spray humidier, and the
panel was placed facedown on a Melinex interleaf. Fine linen canvas
and then hessian (burlap) webbing were placed over the back to form a
moisture-retention layer as well as an evacuation layer, which allowed
a slower drying under a slight vacuum. The procedure, which was con-
tinued for a day with the dehumidier, brought the room RH back to
55% at 21 C. Afterward the panel showed a atter plane.
Under raking light the uneven thinning of the original panel
showed ripples and distortions. Two suitable interleaf materials were
required. After the application of the rst interleaf, the undulations in the
panel were evened out with a ller and then isolated with a second inter-
leaf before attachment to the balsa-wood buildup. A combination of
muslin and then Stabiltex (a very nely woven polyester) was used. Fine
muslin was prestretched on a strainer and coated on both sides with three
coats of Beva 371. The panel was put facedown on Melinex over thick blot-
ting paper on a board on the low-pressure table, and the strainer with the
impregnated muslin was placed over the back. A sheet of silicone Melinex
was placed over the Beva-coated area of the panel, the whole was covered
in Melinex, and a vacuum was applied.
With a heated spatula, the muslin was then bonded to the back
of the panel through the silicone. When it had cooled, the vacuum was
released. Now the panel was attached and could be easily handled on
the strainer.
During these treatments, the table was usually at about 30 C; the
tables built-in dehumidier helped maintain the temperature by control-
ling the RH level. An overall inll of Fine Surface Polylla was applied on
the back of the hessian webbing and sanded at when dry.
A coat of Beva 371 was applied over the leveled layer of Polylla;
a second interleaf was prepared by prestretching Stabiltex on a strainer
and applying three coats of Beva 371 on both sides.
The rst strainer on which the muslin and panel had been
attached was detached. To make sure the painting had adopted a satisfac-
tory surface, it was placed faceup on the board, with webbing under the
muslin up to the edges of the panel, and covered with Melinex. A vacuum
was then applied and the surface observed: the improvement was marked.
For the application of the Stabiltex layer, the painting was laid
facedown on Melinex, and blotting paper and webbing were laid up to the
edges of the panel over the visible edges of the muslin. The new strainer
with the Stabiltex was laid over the painting; silicone was laid over the
panel; and then the whole was covered in Melinex and a vacuum applied.
The Stabiltex layer was then attached with a heated spatula.
The picture was then taken o the table and kept on the strainer in
preparation for the next step, a balsa-wood buildup. Planks of balsa measur-
ing 12.7 63 cm were prepared on the table. In this instance, it was decided
to put two layers of balsa running with the grain of the originalas
opposed to the normal practice of putting the rst with the grain and the
second against the grain. This variation was chosen because it was thought
to reduce slightly the strength of juxtapositioning, as well as to reduce the
chance of any restriction if the panel should move. The balsa wood in this
instance was cut across the grain at 2.5 cm intervals to half of its depth, so
414 Re e ve
that its strength was reduced before application. Both sides of the rst layer
and the underside of the second layer were scored to form a good key.
The panel was cut out of the strainer and placed facedown on
Melinex and blotting paper. Webbing was then placed up to and around
the edges of the original panel. A wooden frame was built up to the com-
bined thickness of the panel plus the rst layer of balsa wood. The pur-
pose of this frame was to reduce the vacuum pressure on the edges so that
there were no distortions in the even downward pressure on the balsa-
wood layer. The heated wax-resin and wood our cement were applied,
and the rst layer of prepared balsa wood put on. The second layer was
415 Srrucrurai Cos trvari o or Pati Pai ri os ar rnt Nari oai Gaiitr, Lono
Fi gure 12
Cosm Tura, The Virgin, reverse. Balsa-wood
buildup on the multipurpose low-pressure
conservation table.
Fi gure 13
Cosm Tura, The Virgin, after panel treatment
and restoration.
applied immediately afterward. The wooden frame was placed around the
edges of the panel; an overall vacuum was then applied and maintained for
a few hours (Fig. 12).
The panel was released and the balsa wood trimmed back to the
edge of the original. The sides were chamfered slightly and the interleaves
turned around and attached by heated spatula to the sides and back. The
sides and back were covered with a ne linen canvas attached by ironing
with wax-resin, and trimmed back to the facing edges. Seen from the side
and end, the panel now has a very slight frontal convex warp. Raking light
photographs show a considerable improvement in the surface.
Subsequently, the holes in the picture were lled and the losses
restored with Paraloid B72; the picture was then varnished with Larapol
K.80 (Fig. 13).
1 The clamping table incorporates longitudinal sash clamps, together with vertical clamping
above and below. All clamps can be moved into any position laterally and vertically. The appa-
ratus has proved to be a great aid in the re-forming and rejoining of panels, especially those
with complex splits, broken joins, and uneven distortions.
2 There is a further reference to the painting in OPD Restauro (1992) (Dunkerton 1993).
Aerolam F board, Ciba-Geigy, Duxford, Cambridge, CB2 4QD, England.
Balsa wood, Solarbo Ltd., Commerce Way, Lancing, Sussex, BN15 8TE, England.
Beva 371, Atlantis European Ltd., 146 Brick Lane, London, E1 6RU, England.
Blotting paper (for humidifying), Arcesso Conservation Materials, 194 Blue House Lane, Oxted,
Surrey, RH8 ODE, England.
Brackets and mirror plates, Frank B. Scragg and Co., 68 Vittoria Street, Birmingham,
B1 3PB, England.
Cascamite (urea-diformaldehyde adhesive and hardener), tool and hardware shops.
Clamps, Buck and Ryan, 101 Tottenham Court Road, London, W1P ODY, England.
Conservation tissue (previously Eltoline, now LX tissue, 100% manila hemp long-ber tissue),
Barcham Green and Co. Ltd., Hayle Mill, Maidstone, Kent, ME 15 6XQ, England.
Evazote, Zotefoams Limited, 675 Mitcham Road, Croydon, Surrey, CR9 3AL, England.
Gator foam, Dixon and Roe Ltd., Units I and II, Bricklayers Arms Estate, Mandela Way, London,
SE1 5SP, England.
Gelatin, Thew Arnott and Co. Ltd., Newman Works, 270 London Road, Wallington, Surrey,
SM6 7DJ, England.
Larapol K.80, BASF United Kingdom Ltd., Dispersions and Pigments Division, P.O. Box 4, Earl
Road, Cheadle Hulme, Cheadle, Cheshire, SK8 6QG, England.
Linen canvas, Ulster Weavers, 47 Lineld Road, Belfast, BT12 5GL, Northern Ireland.
Mastic, damar, and wax, A F Suter and Co. Ltd., Swan Wharf, 60 Dace Road, London,
E3 2NQ, England.
Melinex (Mylar), Preservation Equipment Ltd., Church Road, Shelfanger, Diss, Norfolk, 1P22
2DG, England.
Multipurpose low-pressure conservation tables, clamping tables for panel conser-
vation, spatulas, and irons, Willard Developments, Industrial Estate, Chichester, Sussex
PO19 2TS, England.
Muslin, Russell and Chapple Ltd., 23 Monmouth Street, London, WC2H 9DE, England.
Materials and Suppliers
Notes
416 Re e ve
Paraloid B67 and Paraloid B72, Lascaux Restauro, Alois K. Dethelm A.G., CH 83026 Bruttisellen,
Switzerland; and Atlantis European Ltd., 146 Brick Lane, London, E1 6RU, England.
Paste glue (pearl glue), Brodie and Middleton Ltd., 68 Drury Lane, London, WC2B 5SP, England.
Plastazote, Zotefoams Limited.
Polyester net, John Lewis Partnership, 278-306 Oxford Street, London W1A, England.
Silicone release paper, Custom Coating and Lamination Group, Worcester, MA 01605.
Stabiltex (polyester multilament), Plastok Associates Ltd., 79 Market Street, Birkenhead,
Wirral, Merseyside, L41 6AN, England.
Sturgeon glue, Preservation Equipment Ltd.
Velvet ribbon, Barnett, Lawson, Trimmings Ltd., 1617 Little Portland Street, London,
W1N 5DE, England.
Vinamul 3252 (vinyl ethylene copolymer) dispersion, Atlantis European Ltd., 146 Brick Lane,
London, E1 6RU, England.
Brough, J., and J. Dunkerton
1984 The construction of panel trays for two paintings by the Master of Cappenberg.
National Gallery Technical Bulletin 8:6370.
Brown, C., A. M. Reeve, and M. Wyld
1982 Rubenss The Watering Place. National Gallery Technical Bulletin 6:2739.
Caley, T.
1993 A note on Polylla. Picture Restorer 4 (autumn).
Dunkerton, J.
1993 La Vergine Annunciata di Cosm Tura in Il restauro dei dipinti su ela e tavola:
Problemi ed esperenza. Atti della giornata di studio in occasione dei sessanta anni
del Laboratorio Fiorentino di Restauro dei Dipinti, Firenze, 18 dicembre 1992. OPD
Restauro 5:1622, 6566.
Dunkerton, J., and A. Smith
1986 Ercole de Robertiss The Last Supper. National Gallery Technical Bulletin 10:3337.
Reeve, A.M.
1984 A new multipurpose low pressure conservation table for the treatment of paintings.
Studies in Conservation 29:12428.
1990 A new multi-purpose clamping table for the treatment of paintings on wood. Studies in
Conservation 35:16062.
Reeve, A. M., P. Ackroyd, and A. Stephenson Wright
1988 The multi-purpose low pressure conservation table. National Gallery Technical
Bulletin 12:415.
Smith, A., A. M. Reeve, and A. Ashok
1981 Francesco del Cossas S Vincent Ferrer. National Gallery Technical Bulletin 5:4557.
Wyld, M., and J. Dunkerton
1985 The transfer of Chinas The Incredulity of S. Thomas. National Gallery Technical
Bulletin 9:3859.
References
417 Srrucrurai Cos trvari o or Pati Pai ri os ar rnt Nari oai Gaiitr, Lono
B
rtaxs i rati rai ri os frequently require conservation
treatment. Generally, panel paintings are rejoined to improve the
integrity of the image while preserving the object as a whole.
Common problems include joint failure, splits, and a perceived necessity
to improve joint alignment. Also, the support may need to be strengthened
to forestall deterioration or to prevent the need for reinforcement by other
means that may prove more damaging in the long term. In some cases, the
option of not rejoining may be preferable.
Though a specialized approach may be recommended for the
rejoining
1
of panels, it is not always possible. The following discussion and
outline of general considerations might prove helpful in cases where a
conservator who seldom encounters the necessity of rejoining nds there
are no other means available. Three cases exemplifying panel rejoining
methods used at the Hamilton Kerr Institute (HKI) are described. Each
case represents a particular rejoining problem and the specic treatment
methods and apparatus employed.
Detailed descriptions of rejoining procedures are not common in
conservation literature, although various types of apparatus have been
mentioned (Hermesdorf 1953; Kozlowski 1962; Glatigny 1989; Reeve
1989). Discussion in the section below entitled Smaller Apparatus gives a
basic rationale while providing foundation information for the following
sections. Finally, some disadvantages of the last apparatus, for treating
panels vertically, are discussed. Better methods are continually evolving,
so those described should not be taken as a xed approach.
The pressures applied to rejoin a panel may be divided into two basic types
according to their purpose and orientation in space (Fig. 1). The rst type,
joining pressure, as referred to in this article, is usually directed from the
opposite edges of the panel, and thence, roughly, through the panels
plane and perpendicular to the line of the intended joint. This is usually
done with bar clamps, though other possibilities exist, such as windlass-
type straps, air or hydraulic pressure, and other mechanical devices. The
use of bar clamps to rejoin a signicantly warped panel can make pressure
application dicult. Therefore, it is generally not an ideal method. The
panel may bend into a greater warp, risking breakage, damaging the con-
tact area of the joining surfaces, and negating careful alignment.
Joining Pressure
418
Al Brewer
Some Rejoining Methods for Panel Paintings
3
3
1
2
1
2
Fi gure 1
General direction of application for the types
of pressures (indicated by arrows) used to
rejoin panel paintings: (1) joining pressure,
(2) out-of-plane alignment pressure, and
(3) in-plane alignment pressure.
The second type of pressure, alignment pressure, may be sub-
divided into two categories. Out-of-plane alignment describes pressure
applied roughly perpendicular to the general plane of the panel to bring
the two sides of the joint to the same level.
2
In-plane alignment describes
pressure applied parallel to the joint axis, primarily to bring the elements
of the image into register on either side of a complete disjoin. In-plane
alignment can usually be achieved by maintaining the position of the two
panel members carefully by hand during the rejoining procedure. More
control may be necessary with smooth-faced disjoins, where slippage is
more likely under pressure.
The amount of joining pressure required is determined by the
panels condition. In most cases, much less pressure is used than would
be needed for a construction joint. Panel paintings do not require high
pressures: pressure should be just enough to bring the joint faces snugly
together. If correctly chosen and applied, the glue lls slight gap variations.
Too much pressure is dangerous: it can distort the panel and joint, increas-
ing the possibility of damage to the paint and the structure of the wood.
In fact, some conservators prefer not to apply any pressure during
rejoining to avoid initiating stress. Of course, depending on environmental
conditions following treatment, joints made without the application of
pressure still undergo some internal stress. The use of pressure may also
be defended for the following reasons: (a) pressure can be benecial to a
good glue bond, and (b) a poorly aligned joint is usually dicult to putty
and retouch satisfactorily, especially when a panel painting has a pristine,
glossy surface. Therefore, the application of modest pressure to achieve a
better joint and alignment may be worth considering.
Various systems of wedges and screws with pressures borne by rigid beams
have been developed to control alignment pressures. Weights can consist
of loosely bagged sand or metal pellets, for example. With practice, such
methods can be used with considerable success, though there are usually
drawbacks. For one thing, the bulkiness of some apparatus interferes with
access and control. Moreover, the careful setting of wedges can be frustrat-
ing and tedious and cannot be quickly and easily reproduced if the panel
members need to be moved prior to gluing. Sophisticated, ready-made join-
ing tables that address many such problems, however, can be purchased.
Another approach to rejoining uses (usually) V-shaped wooden
inserts that are glued into channels cut along the line of splits or disjoins.
3
This method will not be discussed here in detail (see Bergeon 1990; Uzielli
and Casazza 1994; see also Uzielli, Historical Overview, and Rothe and
Marussich, Florentine Structural Stabilization Techniques, herein).
The rejoining procedure is often technically demanding. For example,
although there is a choice of adhesives that vary in ease of reversibility, the
diculties inherent in reversing a dried joint usually involve considerable
risk to the structure of the painting, making it desirable to get it right
the rst time. For this reason, control and access are important.
Even the simplest rejoining cases may prove stressful to practition-
ersthis author being no exception. The critical nature of the procedure
demands a purposeful, well-planned approach, the necessity of which can
Precautions and
Suggestions
Approaches to Rejoining
419 Soxt Rtj oi i o Mtrnons ror Pati Pai ri os
become immediately apparent after the glue has been applied and
the joint brought togethera moment when the unforeseen tends to
occur. Contingency measures should be planned beforehand. It is impor-
tant to rehearse the procedure dry (without glue) up to the stage of
pressure application.
It is also important to consider how well a paintings condition can
accommodate the rejoining procedure. Relevant factors are the condition
of the ground and paint layers, whether the layers tend toward aking,
the solubility and reactivity of the adhesive and its components, and the
woods strength and degree of warp. Weak, porous, water-based animal-
glue grounds, for example, might distort or ake during manipulation.
4
A panel can sometimes be pressured into alignment, but inherent
weaknesses could initiate further splits immediately or in the future. The
type of panel wood is an important factor. The more exible woods, such
as poplar, may accommodate greater distortions from pressure without
failing.
5
Accepting less than perfect alignment may be the best alternative
if further treatment might overstress the panel and painting.
6
Gluing procedure varies from case to case. Generally, old glue is
thoroughly cleaned from complete disjoins, which are then aligned and
separated slightly. After glue is applied to both joint faces, the joint is
pressed together with relatively low pressure. For more highly concen-
trated glues, the glue line may be thinned by rubbing (slightly moving
one joint face back and forth against the other by hand or by small
repeated turns of the clamps used to apply out-of-plane alignment pres-
sure). One cannot usually produce a true rubbed joint because the joint
edges would probably cease to move at a moment when the panel is in
the wrong position. However, short of this, a thinner glue linedesirable
for durability and a better match to the original jointcan be achieved.
As splits must be positioned with greater care, rubbing is normally not
possible. For splits, the closest joint is achieved by tting the torn wood
together exactly.
It is not necessary to replane joint faces to eliminate gaps, though
some panels have been so treated. Inserts or gap llers can be used instead.
A replaned joint may be suspected if the image no longer registers where
it crosses the joint. To identify and then treat this condition eectively, it is
bestprior to structural workto remove the varnish, retouchings, and
putties that obscure the joint.
Where joint gaps do occur, llers may be employed; these may be
wooden inserts or part of the adhesive system. If there is an excessive gap
and wooden inserts can be tted eectively without the removal of origi-
nal wood, they are the preferred choice because they use a thin glue line,
which increases durability. Thinner glue lines are more exible and there-
fore able to move with the surrounding wood. In contrast, a glue-saturated
lling compound is more likely to force the surrounding panel to comply
under stresses.
Rejoining and gap lling of joints must be considered in conjunc-
tion with preservation of the original panel wood. Some conservators pre-
fer to replace the original wood with wooden inserts, usually V-shaped in
section, whose good t should result in a more complete and thinner bond
line than that achieved by rejoining the original wood unaltered. The
joints (for two new joints are created) can be made as sound as technical
skill, patience, and materials will allow. Again, because the glue line can
420 Bre we r
be made thin and, therefore, more exible, joint strength and durability
should be better.
However, if sucient strength can still be achieved, it may be
preferable to leave the original wood intact at a disjoin or split to preserve
its established relationship with the painted side. Many breaks can be
rejoined adequately without removal of the original panel wood. If a
panel breaks again in the same area, the original wood can still be repaired
or even, as a last resort, replaced. Compromise may be required when
insect damage is a factor. In any case, it is probably better to avoid or
minimize the loss of original wood support.
Longer joints are dicult to rejoin in one procedure. Glues have
limited open times, during which they are suciently liquid to allow
eective manipulation. With a larger joint, a step-by-step closure may be
advised. The use of insert methods would allow this possibility. The choice
of a method, or a combination of methods, is a question of judgment.
Access to both sides of a panel, especially the painted side, is desirable in
order to assess the eects of the procedure, promote easy glue applica-
tion and removal, judge the relative position and angle of the two parts
being joined, control the degree and direction of pressure for alignment
and rejoining, and allow the placement of pressure where it will be
most eective.
There are disadvantages in having access to the back onlya limi-
tation that can occur, for example, when the panel is treated facedown on
a table surface. The primary drawback is that it is impossible to judge the
alignment of the paint surface because it is not visible. This is especially
important if the painting has been previously misaligned and the panel
subsequently thinned, because the plane of the back surface cannot be
relied upon to ensure realignment of the plane of the painted side. The
original paint surface usually provides the best basis for alignment.
Access to the true, original paint surface is desirable so that the
paintings integrity can be respected during the procedure. Old putties
may have been imperceptibly ramped to disguise previously misaligned
joints so that neither local alignment nor the general plane of the painting
surface can be judged with accuracy.
7
Judgment of the general plane is a
particularly subtle exercise that demands thorough familiarity with the
panels surface conformation.
In addition, overlying nonoriginal layers (i.e., putty fragments
falling into the joint) can obstruct closure. This usually occurs when all
other preparations have been made and the glue has been applied. If a par-
tial disjoin is bridged by such layers and disjoins further during treatment,
then original paint on either side of the joint may stick to the overlying
layers and be dislodged.
This article describes three types of apparatus used by the author at
HKI to glue disjoined or split panel paintings. One is relatively simple in
construction and suited to smaller panels. The other two were built for
larger panels.
One advantage of the rst type is its ease of quick assembly and
disassembly. The other two types are more elaborate structures, but they
Apparatus for
Rejoining Panels
Access and Preparation
421 Soxt Rtj oi i o Mtrnons ror Pati Pai ri os
can be taken apart and rebuilt to suit most larger panels or be customized
for a particular situation. All three designs require a degree of thought and
planning in their application. However, they are relatively inexpensive,
given the control and exibility they allow in the gluing process.
All of the designs utilize a type of screw clamp, sometimes known
as a hold-down clamp, to provide pressure (Fig. 2).
8
The screw clamp is
mounted on a suciently rigid beam, usually of right-angled-section metal
that is xed in relation to the panel.
9
The spatial arrangement of the clamp
and beam determines the general direction of pressure. The clamps are
used primarily to achieve the desired alignment of joints in relation to the
general plane of the panel, that is, to reduce steps. They can also be used
instead of bar clamps to provide joining pressurefor example, where
greater directional control is desired.
The screw clamp can be attached to any suitably thick piece of
stock. The thumbscrew of the attachment device may be snugged securely
in position with pliers. The clamps are small enough to be placed closely
together, and they can be moved to any desired location along the mount-
ing beam. The screw shown can be adjusted through a length of about
20 cm. The circular swivel foot piece can be modied by padding or by the
attachment of shaped pieces with various contact areas and rigidities in
order to spread the applied pressure as desired.
Case description
A seventeenth-century panel painting
10
was treated structurally for splits
from a cradle locked by glue that could not accommodate the paintings
response to environmental uctuations. The panel consists of two planks
joined parallel to the grain near the center. The grain is oriented vertically
with respect to the image. Two splits had occurred since cradling, shown
by the lack of glue or varnish in the splits. The splits were stepped to a
small degree.
Order of rejoining
The panel, which was almost as thick as it had been originally, had been
cradled unnecessarily. The cradle was removed to permit access for rejoin-
ing and to serve as a preventive measure against further splitting. The
extent of splitting was small, with the splits closed at both ends.
Smaller Apparatus
422 Bre we r
Fi gure 2
Screw clamps. One is attached to a double-
thickness length of right-angled-section metal.
In panels with multiple splits, some running the entire length of
the panel, it is preferable to rejoin each section rst. This is partly because it
becomes more dicult to control the procedure as the size of each section
increases. Joining pressures must be directed over increasingly greater spans,
and sections with unconsolidated weaknesses, especially larger sections, are
more dicult to manipulate than those that have been consolidated.
Apparatus description and application
Construction
The apparatus is supported by a table frame with crossbars (Figs. 3, 4a, b).
In order of assembly, a single alignment frame is made rst from two
equal lengths of right-angled-section aluminum, for lightness and sucient
rigidity. The aluminum lengths should be cut at least 50 mm longer than
the dimension of the panel that is parallel to the intended joint. The
lengths, which determine the maximum size of panel that can be treated,
are drilled at each end and bolted together with two shorter lengths of at
metal to make a rectangular frame. Two such frames, one for each side of
the joint, may be necessary to achieve sucient control of joint alignment
perpendicular to the panel plane.
423 Soxt Rtj oi i o Mtrnons ror Pati Pai ri os
Flat
metal
4
Panel
C-clamps
Table
frame Bar clamps
Threaded
rods
Wooden
beams
Right-angled-section
metal
1
Alignment
frames
3
2
Fi gure 3
View of the smaller rejoining apparatus, in
which the order of assembly is as follows:
(1) build and attach the alignment frame(s)
to the table crossbars; (2) place two parallel
wooden beams on either side; (3) join the bar
clamps with threaded rods and place on the
wooden beams, through the alignment
frame(s); (4) position the panel and adjust sup-
ports, clamp positions, and pressures. (To sim-
plify the diagram, only one panel member is
shown in position, and the bar-clamp stops
are not padded.)
Fi gure 4a, b
Cornelius Janssens (or Johnson), Portrait of the
Third Earl of Moray, seventeenth century. Oil
on oak panel, 807 640 7 mm thick.
Private collection, Scotland. HKI treatment
no. 1475. A view (a) of the rejoining proce-
dure shown from above an end-grain edge.
Note the restraining bars to the left and right
of the alignment frames, attached with
C-clamps to spacer blocks and the table frame
below. A view (b) from underneath the mid-
dle of the panel shows the alignment frames
and six screw clamps xed to the table cross-
bar with C-clamps.
a b
The screw clamps will be attached to the angle-sectioned beams
above and below the panel joint so that they are on either side and in line
with it. First, however, the alignment apparatus is positioned approxi-
mately and the bottom beam clamped to the table crossbars with small
C-clamps, which stabilize the apparatus.
Next, two straight wooden beams, of about 50 50 mm in cross
section, are placed on each side of the alignment frame(s). These may be
clamped to the table crossbars. Then, depending on the panel size, at least
two bar clamps are laid across the wooden beams and through the rectan-
gular frames. The beams above the lower set of screw clamps support the
bar clamps and the panel. The top surface of the bar-clamp rail should lie
in the middle of the alignment frame(s). This arrangement denes the
panel position in relation to the clamps.
For stability, all the bar clamps may be joined by some relatively
rigid means so that they are parallel to one another. In the diagram, two
standard threaded steel rods serve the purpose, passing through stop holes
and xed with nuts to either side of each bar clamp. Fixing the clamps
together rigidly prevents accidental slips and provides a secure base for the
panel. Depending on the panel shape and the angle of the joint relative to
the panel edges, the clamps can be positioned at angles to the panel edges
rather than placed strictly perpendicularly.
Prior to placement of the panel in the apparatus, the eective
contact area of the stops of the bar clamps is extended and padded. A
length of relatively rigid bar (e.g., a strip of wood) is placed against the
line of bar-clamp stops at each panel edge. A thin balsa plank or strip of
card is placed between the rigid bar and the panel edge.
These two pieces distribute the pressure more evenly along the
entire panel edge. The batten spreads the point pressures of the stops, and
the padding conforms to local irregularities. The padding material can be
carved or sectioned to apply pressure to the strongest surface while it
avoids weaker areas. The lengths of batten and padding are cut slightly
shorter than the respective panel edge. To permit judgment of curvature
during the procedure, they are positioned to allow sighting along the end-
grain edge of the panel.
Panel manipulation before rejoining
Before glue is applied, a dry rehearsal of the alignment procedure is con-
ducted. To bring both sides together squarely, it is critical to respect the
panels curvature during rejoining. Otherwise, a poor joint usually results,
with interruptions of the inherent contours of the panel surface at the joint.
If the panel is weak or warped, it should be supported in a state of curva-
ture that minimizes the bending stresses imposed by its own weight. This
can be done by placing wooden shims at intervals beneath the panel which
are cut to ll the gap between the panel back and the bar clamps (Fig. 5).
The panel is then slid horizontally, painted side up, onto the bar
clamps and through the rectangular void of the alignment apparatus until
the intended joint is approximately aligned with the line of screw clamps.
Convex warps (viewed from the painted side) often promote buck-
ling when joining pressure is applied. Inherent warp and excessive side
pressure increase the tendency to buckle. This pressure can be redirected
through the panel toward the desired direction and across the intended
joint by the positioning of restraining bars above the panel.
424 Bre we r
a
b
Panel
section
Restraining
bars
Shims Liquid
glue
Fi gure 5a, b
Views of the end-grain edge of a warped
panel, with the curvature supported by shims
from below. Under joining pressure (indicated
by arrows), when the panel is unrestrained
(from above) from buckling (a), the joint
aligns poorly and a gap is created between the
joint faces, which will not be well bridged
when the glue shrinks and dries. Padded
restraining bars above the panel surface
(b) redirect the joining pressure through the
joint. The joint faces then meet more squarely
and squeeze glue evenly from above and
below, indicating a better joint conguration.
In general, the restraining bars are placed parallel to and approxi-
mately halfway between the joint and each parallel panel edge. The bars
may be made of padded wooden strips of sucient rigidity. They should
barely contact the panel surface or lie slightly above it. Spacer blocks are
placed beneath each end of the bars, and the ends are clamped.
If necessary, the alignment frame is repositioned, after which the
screw clamps are lightly snugged to the panel surface. The foot pieces are
isolated from the panel and glue with a release lm
11
and padded, if neces-
sary, with small pieces of mount card or blotting paper. Glue should not
contact the card or paper, as it could seep beneath and damage the painting.
Out-of-plane alignment of the joint edges is usually best attained
with the least number of screw clamps and with the least pressure applied
at the least number of pressure points. The conservator determines the
arrangement by trial, repositioning the clamps until the desired eect is
achieved. The procedure is usually to move each joint edge alternately, and
about equally, until alignment is achieved. One edge should not assume all
the strain. Many splits and disjoins realign with ease when the simplest
appropriate arrangement of pressure points is used.
Out-of-plane alignment and the best overall curvature may be
determined by several methods. These include (1) passing of the nger-
tips across the joint, (2) repeated passing of the palm across the general
area of the joint, (3) use of raking light cast across the joint from both
sides, (4) sighting of panel edges at the ends of joints (if appropriate),
(5) checking the gap with backlight, and (6) use of raking light or back-
light, with straight edges placed over the joint.
During use of these techniques, the joining pressure is tested, a
process that previews how the panel shape will change under the antici-
pated pressure. Alignment pressure may have to be adjusted slightly in
accordance with a shape change, and further precautions may be neces-
sary. For example, thinner panels may bend in plane when joining pressure
is applied to a joint that is gapped in the middle (Fig. 6). The joint edges
contact near each end while the gap is reduced. This type of bending
increases as joining pressure is concentrated across the gap. It may occur if
the padded bars are not suciently rigida deciency that causes pressure
concentrations where the bar-clamp stops make contact.
To control these eects, it may be necessary to shim the curved
edges of the panel (Fig. 7). Very small movement can have a signicant
eect on nal alignment and bond strength. Shims can be used to concen-
trate joining pressure to close or reduce slight joint gaps.
Glue application and rejoining
After successful completion of the dry rehearsal discussed above, the
conservator can proceed with the application of glue and the actual rejoin-
ing. To allow access to the joint for gluing, the top right-angled-section
beam(s) and screw clamp(s) may be entirely removed from above the
panel, or a bolt may be removed from one end only and the beam(s)
hinged up and away. Alternatively, each top screw clamp could be backed
o the panela maneuver that may be preferable and wastes little time
during repositioning after or during glue application. The bottom clamps
provide a suciently xed datum if the panel is relocated exactly. Another
option would be to mark each screws position with an ink line across the
screw thread and screw housing, back o the clamp, apply the glue, and
425 Soxt Rtj oi i o Mtrnons ror Pati Pai ri os
Fi gure 6
The eect of even joining pressure (indicated
by arrows) on a disjoin with a central gap
(exaggerated). The dotted lines show move-
ment of the panel under pressure as it bends
in-plane.
Shims
Panel
Wooden beam
Fi gure 7
Shims placed in gaps along an uneven edge,
left, spread the joining pressure (indicated by
arrows) evenly along the joint.
turn the screw back to the mark. For most joining procedures, any disrup-
tion of the panels position relative to the screw clamps necessitates minor
readjustment when the joint is nally brought together.
Depending on the joint, the glue can be applied with the entire
panel removed from the apparatus, or, if the joint consists of two pieces,
one piece can be positioned and clamped in place while the other piece is
moved slightly away to create a sucient gap. The open structure of the
apparatus allows considerable access for brushes and ngers.
Next, the glue is applied. Care and ingenuity must be used for
partial disjoins, especially splits with both ends closed. An excess of glue
is worked into the break, preferably from the panel back. Methods to
increase glue penetration include nger pressure, slight exing of the joint
edges, suction, positive air pressure, and use of a syringe or a spatula. For
better wetting of the joint faces, a more dilute glue may be applied rst,
then a more concentrated glue. The highest practical concentration should
be used to avoid starving the joint,
12
a condition that can occur with
glues that shrink or dry by moisture or solvent loss or with glues that are
partially absorbed by porous woods.
To produce as complete a joint as possible, it is sometimes better
to leave a sucient line of excess glue on the back of the panel only,
since glues that dissolve or disperse in water or solvent usually shrink
into the joint. Any outstanding dried glue is then removed to the level of
the panel surface.
The clamps are reset, and joining pressure is applied lightly in
small increments, with alignment readjusted if the joint slips. The aim is
to maintain alignment while forcing excess glue out of the joint in equal
measure from the front and back of the panel. This indicates that the join-
ing faces are meeting squarely.
If joining pressure is directed nearer to the front or back of the
panel, a gap may result toward the opposite side. This occurs, for example,
when a buckling deection is induced in a panel with an inherent warp
(Fig. 5). The chances of making a starved joint can be reduced by a slight
increase in overall pressure in two or three stages during the initial drying
period. In this way, shrinkage and absorption of the glue are countered by
a reduction of the joint gap.
It may be necessary to readjust the alignment screws intermit-
tently between successive increases of joining pressure. This is especially
true for thin, exible panels and for disjoins, where movements are more
likely. Disjoins, because they are usually straight and smooth, are often
prone to slippage as joining pressure is applied (Fig. 8). Joint slippage
can occur imperceptibly, long after the nal pressure settings have been
made and well into the initial setting stages of the glue. It is necessary
to check the alignment repeatedly in all directions until the glue is set to
ensure best results.
After the glue has dried, pressure mechanisms are released in the
order opposite to which they were applied during the gluing procedure
(rst the bar clamps, then the alignment screws). The bar clamps are
backed o in small increments, in the order and to the degree in which
they were applied. Any unexpected movements or sounds may signal a
critical weakness. If the alignment screws are released before the bar
clamps, then critical support may be removed prematurely from the joint
area, and the panel may buckle.
426 Bre we r
Fi gure 8
Slippage movement (indicated by vertical
arrows) induced under joining pressure (indi-
cated by horizontal arrows). This type of slip-
page occurs when the axis of the joint is not
parallel to the panel edge(s) and therefore is
not perpendicular to the applied pressure.
Case description
This apparatus was constructed in 1988. The method follows the principles
described in the previous section. Screw clamps are arranged by some
means around the panel to provide joining pressure. Alignment pressure is
applied perpendicularly to the panel plane.
A more extensively damaged panel
13
of larger dimensions than the
one discussed above necessitated construction of a more versatile combina-
tion of support table and rejoining apparatus. It was necessary to remove
battens to gain access to splits and to insect-damaged wood. The panel
required interim support until all the splits were glued and an auxiliary rein-
forcement applied. As the restraining battens were removed and the splits
glued, it was expected that the panel conformation would vary accordingly,
so that the interim support would have to be made adjustable to panel
warp. Again, right-angled-section girders were used for construction.
Apparatus description and application
A main table was constructed that could, as work proceeded, support the
panels changing curvature across the grain direction and also provide a
framework from which joining apparatus could be applied (Fig. 9a, b). The
Larger Apparatus
for Treating Panels
Horizontally
427 Soxt Rtj oi i o Mtrnons ror Pati Pai ri os
Legs
Feet
Right-angled-
section metal
Wooden
beam
Soft foam Panel Blockboard
Extension Box-frame table
Fi gure 9a, b
Drawing of an apparatus for rejoining larger
panels, consisting of a table with extension,
shown in elevation (a) from the end-grain
edge of the panel. Note the angle adjustment
of the extension, which accommodates the
panels curvature, and the turnscrews on the
left, which are angled so as to direct joining
pressure through the panel plane, thereby
reducing buckling tendency. A view from
below (b) shows the same end of a large panel
during the rejoining of a split. The panel is
facedown, and the facing is removed only in
the area surrounding the split.
a
b
table consisted of a box frame of metal girders supporting a top of 25 mm
(1-in.) thick blockboard panel fastened with screws from below. A layer of
soft, 15 mm thick polyurethane foam sheeting
14
was laid on top, and a
release lm was secured over the foam.
The main table was extended to accommodate the panel and join-
ing apparatus during gluing procedures. A side edge of the panel was pro-
jected over the table edge to expose the split being treated at both the
front and the back of the panel. The panels projected edge was supported
by an extension, separated from the main table with a variable gap. The
gap allowed access from above and below to the area to be glued. Girders,
nested double for sucient rigidity, were attached to the table frame.
Screw clamps could be positioned to apply pressure in any direction for
alignment and rejoining.
As work progressed toward the center of the panel, it became
necessary to project an increasingly large portion of the painting, sup-
ported by the table extension. The extension consisted of a padded panel
lying on four upturned screw clamps, which were attached to the girders
of the extension frame. The extension girders, in turn, were attached to
and extended from beneath the main table. They were of sucient length
to double the main tables width when fully projected.
The padded extension panel was thus made adjustable for angle,
height, and distance with respect to the main table. These factors permit-
ted adjustment of the panels plane to conform with varying warp or to
achieve various angles for gluing. Eventually, as more of the painting was
projected, it became necessary to reinforce the projected girder ends with
footed vertical girder legs that rested on the oor.
As rejoining proceeded, an inherent convex warp
15
became appar-
ent when the panel was viewed from the painted side. The legs could be
angled to direct the joining pressure in order to align it with the panel
warp. Because the panel was facedown, pressure was directed at a slightly
downward angle, in line with the panels curvature, to prevent buckling.
Batten removal and rejoining began from one side-grain edge and
was continued toward the center.
16
After half of the panel was consoli-
dated, it was turned 180 horizontally to treat damages to the other half.
For each split the battens were removed from above and to a point just
before the next split. The exposed split was then aligned and glued.
Splits occurred at various angles in relation to the panel edge and
roughly parallel to the local grain. The direction of any split could be fol-
lowed closely by the screw-clamp positions, since the girders to which they
were attached could be bolted at any angle in the horizontal plane. The
top girder(s), with clamps attached and set, could be unbolted at one end
and pivoted away from the split for the application of the glue, then repo-
sitioned quickly for the application of joining pressuremuch as was done
with the smaller apparatus.
In such a large rejoining mechanism, the beams that support the
alignment apparatus are often not suciently rigid, especially when pres-
sure must be applied in the middle of a large panel. Rigidity may be
increased by bolting two lengths of girder together in the most useful
conguration. Nested T or U sections may be constructed. U sections will
allow screw clamps to be placed in parallel lines. Any thickness of timber
could also be screwed or bolted to a girder to increase rigidity.
The entire apparatus can be taken apart quickly and easily, and
the parts can be stored in a relatively small space or used for another pur-
428 Bre we r
pose. Several other modications can be made to the system, depending
on need and limited only by imagination.
Case description
The treatment of a large eighteenth-century panel
17
suggested another
rejoining method. The panel had been thinned. Rigid steel edge strips had
been screwed into the end-grain edges of the horizontal planks, preventing
movement of the panel across the grain during humidity changes. The
resulting constraint caused considerable disjoining, partly because of poor
environmental control.
The panel could have been treated horizontally, as in the case
described above. However, a more compact apparatus was used to provide
access and to make ecient use of studio space. Vertical orientation of
the painting is also advantageous because it allows easier access to both
sides than if the panel were oriented horizontally. Another benet is that
some aspects of cleaning, lling, and retouching can be conducted in tan-
dem with the structural work if both sides of the panel are almost com-
pletely exposed.
Apparatus description and application
A frame/trolley constructed of a metal girder with six wheels for mobility
was converted to a temporary support during treatment.
18
A padded ledge
was axed to the trolley bed, and the panels longest edge was laid onto it,
so that the panel planks were vertical with their backs facing outward.
Two silicone paper strips placed beneath the lower panel edge reduced
friction to allow warp movements. The topmost edge (a side edge of the
panel) was supported with a padded length of girder. Thus, the panel was
positioned for rejoining, back outward, as if on an easel.
Joining pressure was applied with polyester webbing straps tted
with ratchet-uptake mechanisms. This type of strap, available in various
lengths, is typically used to tie down loads for haulage. The principle is
similar to that of a windlass-type tourniquet. Such tourniquets can be
manipulated to create a greater variety of pressure options than are pos-
sible with bar clamps.
Two straps (rather than one) were used. They were joined end to
end to encircle the panela method that achieved one line of pressure.
This method was used because the ratchet mechanism, if it were located
on only one side of the panel, would cause unequal tension on each side
because of friction at each end of the panel due to strap pressures. The
resulting constraint produces a bending pressure toward the uptake side.
However, when there is a ratchet on each side, pressure can be applied
equally or unequally, as desired.
The joining pressure of the straps was applied to the panel edges
through rigid end blocks, made from lengths of padded wood bolted into
girder lengths. At each end of the panel, a strap was run through a slot
in the girder and around the girder and the outside of the block. Each
ratchet was loosely suspended from such an end block, then positioned to
bear against the block when pressure was applied. Slings of cord or web-
bing were used to suspend the end blocks and ratchets from the top retain-
ing bar of the support frame, where the bar projected beyond the borders
of the painting. Thus the line of pressure could be directed at any desired
Larger Apparatus for
Treating Panels Vertically
429 Soxt Rtj oi i o Mtrnons ror Pati Pai ri os
height. The slings also prevented accidental damage that could have
resulted if the end blocks had contacted the panel painting.
For longer splits or disjoins, it was necessary to use two or more
strap pairs to concentrate the pressure across the entire panel. The num-
ber and the location of straps and the lengths of the end blocks deter-
mined the location and distribution of pressures.
Alignment was achieved with screw clamps and girder lengths,
as discussed for the previous case. In this case the girders were placed on
either side of the panel with their longest axes vertical and parallel to the
joints or splits. The clamps could be repositioned to adjust alignment.
Then, the girders could be unbolted at the bottom and pivoted away on
either side of the panel to provide access for glue application. The bars
could then be rebolted in virtually the same position, with only slight
adjustments to the alignment clamps being necessary.
Each disjoin was treated consecutively across the panel. As work
progressed, the girders and clamps were moved to the adjacent disjoin.
There were major disadvantages to the vertical apparatus. For one
thing, access to the lower edge was limiteda problem that could be over-
come by an improvement in design. Moreover, if a painting is especially
heavy and if movement (from changes in moisture content, for example)
occurs during treatment, the resulting friction would impose constraint.
Another drawback of the vertical apparatus is that gravity can adversely
aect the ow of adhesives and consolidants. The vertical orientation can
make it dicult to control tools in such procedures as using chisels to t
wooden inserts into areas of severe insect damage, especially toward the
lower edge. One nal caution is that the vertical orientation can be used
only in cases where the paint is secure or well faced, or there may be
losses due to aking.
Most described treatments were done while the author was an intern spe-
cializing in panel painting conservation at the the Hamilton Kerr Institute
(HKI). Thanks go to the Getty Grant Program and to the Samuel H. Kress
Foundation, New York, for funding the internship. Other treatments were
completed while the author continued at the HKI, employed as a conser-
vator and research associate, thanks to funding by the Leverhulme Trust,
London, and the Samuel H. Kress Foundation and the HKI. The author
thanks Ian McClure for his support and, above all, for allowing him free-
dom in pursuing these treatments.
1 In this text the term rejoining refers to the gluing of either splits in the wood support or joints
that have failed, or disjoined, due to a glue line being too weak or deteriorated. The term joint
is used more generally and refers to the line where two wood members meet or would meet,
whether the joint is intact, disjoined, or split.
2 The joint edges are displaced such that one edge is above and the other is below the general
plane of the panel. Such misalignment is sometimes called a step or stepping.
3 The method described in this article utilizes specialized apparatus to rejoin a particular break
in a single gluing procedure. In contrast, the insert method usually rejoins segments of a dis-
join in sequential steps so that the joint is treated with successive gluing procedures. The
insert method generally avoids the use of joining pressure as dened in this article.
Notes
Acknowledgments
430 Bre we r
4 Generally speaking, most water-based glues used for rejoining are removed easily from a paint
surface while they are still moist. It is important to remove them as soon as possible after they
contact the paint, since paint losses can result from swelling or solvent eects. Also, strong
hygroscopic glues that expand and contract, such as animal (collagen) glues, though soluble
when dry, can easily detach underlying paint (see Mecklenburg 1982:gs. 9, 11). Because even
a thin remnant of such a glue can detach paint, it should be thoroughly and immediately
removed from the paint around the joint.
5 This characteristic has probably been the salvation of many poplar panels subjected to stress-
ful framing and reinforcement structures.
6 Like many conservation considerations, the determination of what constitutes overstress in
the manipulation of a panel painting is generally a matter of judgment, based on experience
and common sense.
7 Rather than leaving one side of a joint higher than the other, an earlier conservation might
have graded a putty or ller between the two levels. Such a grade, or ramp, is often a sign of
inaccurate rejoining or of a break that has been supercially treated without structural work.
8 The screw clamps potential was suggested to the author by Professor I. S. Hodkinson of
Queens University, Kingston, Canada, where it was applied for this purpose. The rst appara-
tus used at HKI was designed and built in winter 1987. See Materials and Suppliers below, for
the supplier of the hold-down clamps used in this apparatus.
9 This is also known as slotted angle and is found in various forms in laboratories in many
countries.
10 Cornelius Janssens (or Johnson). Portrait of the Third Earl of Moray, seventeenth century. Oil on
oak panel, 807 640 7 mm thick. Private collection, Scotland. HKI treatment no. 1475.
11 Polyester (polyethylene terephthalate) lm.
12 A condition in which insucient glue remains in the joint after drying.
13 Marco Palmezzano, The Mystic Marriage of Saint Catherine, 1537. Oil and egg tempera on
poplar (visual identication) panel, 2560 1805 20 mm thick. Signed and dated. Property of
the Marquess of Northampton. HKI treatment no. 1302. (See Brewer, Practical Aspects,
herein, and Brewer 1994.)
14 The foam was used as a thin padding to distribute the weight of the panel. Use of a hard
surface, which would have concentrated the weight on too few points over the warped panel
surface, would have risked damage to the paint during the relatively long treatment period.
15 A curvature typical of substantially thinned panels.
16 See the authors discussion of the removal of reinforcements from large panels (Brewer,
Practical Aspects, herein).
17 Anton Raphael Mengs, Noli Me Tangere, 1771. Oil on walnut panel, 2915 1785 20 mm
thick. The Warden and Fellows of All Souls College, Oxford. HKI treatment no. 73.
18 See Brewer, Practical Aspects, Figures 7, 9a, 9c, herein.
Hold-down clamps, Trend Machinery and Cutting Tool Ltd., Unit N, Penfold Works, Imperial
Way, Watford WD2 4YY, England.
Bergeon, S.
1990 Science et patience, ou La restauration des peintures. Paris: Editions de la Runion des
Muses Nationaux.
Brewer, A.
1994 A consolidation/ller system for insect-damaged wood. Hamilton Kerr Institute
Bulletin 2:6872.
References
Materials and Suppliers
431 Soxt Rtj oi i o Mtrnons ror Pati Pai ri os
Glatigny, J-.A.
1989 Evolution des matriaux utiliss lIRPA, Bruxelles, travers un exemple dans le
domaine du collage des panneaux. In Conservation-restauration des biens culturels:
Traitement des supports, travaux interdisciplinaires, 4547. Paris: Association des
Restaurateurs dArt et dArchologie de Formation Universitaire.
Hermesdorf, P. J. F. M.
1953 Joining loose members of panel paintings. Studies in Conservation 1:8791.
Kozlowski, R.
1962 An apparatus for gluing split panels. Studies in Conservation 7(4):13540.
Mecklenburg, M.
1982 Some aspects of the mechanical behaviour of fabric supported paintings. Report,
Smithsonian Institution, Washington, D.C.
Reeve, A. M.
1989 A new multi-purpose clamping table for the treatment of paintings on wood. Studies in
Conservation 35:16062.
Uzielli, L., and O. Casazza
1994 Conservazione dei dipinti su tavola. Fiesole: Nardini Editore.
432 Bre we r
T
ni s arri cit cosi ntrs the problems encountered in the framing
of panel paintings in cases where some movement in response to
uctuations in ambient levels of relative humidity (RH)either in
short-term cycles or longer-term, seasonal cyclesis anticipated. The pan-
els considered here are those that can be handled without further rein-
forcement and that can accommodate some movement without buildup
of stress, as well as panels that have support systems that move with the
panel and require a rigid frame to enclose and protect them. Particularly
sensitive panelsthose that are at risk from conicting tensions in the
structure or those weakened by agents of degradationshould always
receive further structural treatment or climate-controlled enclosures.
In this article the frame itself is regarded as an auxiliary rigid sup-
port. The methods used to construct a frame for a painted panel illustrate
several principles: dierent materials may be employed according to avail-
ability; the panel must be able to expand, contract, and warp in response
to changes in RH; and in some instances, simple, unobtrusive modica-
tions are made to the frame. The systems described here are the result of
experience gained during the tting of panels for display in places where
the environment cannot be precisely controlled, such as in private collec-
tions, or in situations where small, unsupported panels have been prepared
for transport and display in temporary exhibitions.
The rate of response to uctuations in RH will vary depending on
the thickness of the panel, the type of wood, the cut of the planks from
which the panel is made, and the degree of sealing of the reverse of the
painting and the endgrain. The framing should also take into account the
amount of movement the panel is likely to produce within the range of
RH levels in a given environment, and the space for that movement should
be built into the frame rabbet.
Notably, the number of articles that discuss framing panels is rela-
tively small; this situation may reect the fact that framing often falls out-
side the jurisdiction of the conservator of paintings. With the growing
popularity of large-scale traveling exhibitions and with their accompanying
risks, however, it is essential to review and evaluate the principles behind
framing methodsand perhaps arrive at some comprehensive guidelines.
A historical survey of the framing of panels
1
could start with inte-
gral frames, where the frame is carved from the same panel on which the
painting is executed, such as the portrait of Emperor Charles V, attributed
to the Master of the Magdalen Legend and painted at the beginning of the
433
Ian McClure
The Framing of Wooden Panels
sixteenth century (Fig. 1), or a large, complex altarpiece where the mold-
ing of the frame is securely attached, sized, and gessoed along with the
panel. An example is the San Pier Maggiore altarpiece by Jacopo di Cione,
most of which was removed from Florence and is now in the National
Gallery, London.
2
While there is evidence (such as the xing of battens at
only one point on each vertical plank of the painting) that altarpieces were
constructed to allow small movements,
3
it seems likely that the relative sta-
bility of environmental conditions
4
within the church or chapel mitigated
the buildup of tension and stress, which could result in cracking and split-
ting. Elements of the San Pier Maggiore altarpiece, however, were proba-
bly glued, dowelled, and nailed together with battensprocedures that
produced a very rigid structure to counter the artworks size and weight.
In northern Europe, panel and frame construction tended to be
more sophisticated than in the south, and frames were routinely designed
to allow movement of the panel.
5
For example, the wings of the Oxburgh
Altarpiece, produced in an Antwerp workshop around 1530, have the pan-
els tted, unglued, into grooves in the frame molding. Despite allowances
for movement, large altarpieces of this type are known to have suered
from structural failure due to aws in their original construction. For
example, it has been suggested that modications had to be made to the
wings of van Eycks Ghent Altarpiece, as the wings proved to be too
heavy (Verougstraete-Marcq and Van Schoute 1989:78). In the case of the
Oxburgh Altarpiece, structural failure was a result of a restoration that
was based on a misunderstanding of the principles behind the original
construction. The free expansion and contraction of the panel in its frame
had produced a gap between the malrand (paint edge) and the frame edge.
This was lled and retouchedrestorations that proceeded to restrict the
panels movement and cause splits in the panel and tenting and aking in
the paint layer (McClure and Woudhuysen 1994:2023). The rigidity of the
434 Mc Cl ure
Fi gure 1
Master of the Magdalen Legend (attrib.),
Emperor Charles V, early sixteenth century. Oil
on oak panel with integral frame, 34.3 23.8
cm. Fitzwilliam Museum, University of
Cambridge (2309).
frames of the wings was further weakened by the tting of brass bolts and
keeps in the nineteenth century (Fig. 2).
By the mideighteenth century, the movement of paintings from
their ecclesiastical settings into private collections and museums had
begun. Complex altarpieces were broken up and installed in new, fashion-
able frames, losing in the process not only cultural context but also, in
many cases, structural soundness. For example, the context was obscured
in a small Virgin and Child, painted in Florence in the 1420s and now in the
Fitzwilliam Museum, Cambridge, when the arched top was squared o
with a wooden addition decorated with gilded pastiglia work, so as to t
a rectangular frame, presumably for display in a secular setting (Fig. 3).
A portrait of a man by Memling, originally part of a diptych, depicting
a donor and presumably the Virgin and child (part of the Bearsted
Collection, Upton House, National Trust), now has a nineteenth-century
ornate Gothic frame tted inside a shadow box. Traces of the original mal-
rand survive, as do traces of gilding from the original, integral frame.
There seems to be no evidence that panel paintings were ever
tted with a regard for expansion, contraction, and warping of the panel
support before the twentieth century, with the exception of double-sided
elements of altarpieces in northern Europe. Even such a grand altarpiece
as Carlo Crivellis Madonna of the Sparrow, probably commissioned in the
1490s (National Gallery, London), has developed cracks as a result of its
original construction. The altarpiece is largely intact, although the central
panel has been thinned and cradled. The predella panel, a single horizontal
plank painted with three separate scenes, was securely nailed in with nails
435 Tnt Fraxi o or Woont Patis
Fi gure 3
Florentine school, Virgin and Child, early
fteenth century. Tempera and oil on poplar
panel, 84.5 45.2 cm. Fitzwilliam Museum,
University of Cambridge (1987).
Fi gure 2
Antwerp school, Oxburgh Altarpiece (right
wing, outer side), ca. 1530. Oil on oak panel,
226 114 cm. National Trust, Oxburgh Hall,
Norfolk, England. In the lower right corner is
a bolt that was inserted later.
of diering lengths. One nail subsequently caused a horizontal crack
(Smith et al. 1989:32, 37, g. 7).
The method of nailing panels rigidly into frames seems to have
become generally employed as soon as frames were recognized as separate
from the paint support. It is not uncommon to nd panel paintings in
British private collections that have been secured in this way and left
undisturbed for generations. A pair of early-seventeenth-century portraits
of Edward Altham and Elizabeth Altham, from Kingston Lacy House in
Dorset, England, now owned by the National Trust, survive in their origi-
nal frames and have received little, if any, structural conservation mea-
sures. Pinned tightly at regular intervals around the edges, the panels, each
formed of three vertical oak planks, have been unable to move in response
to changing levels of RH. A joint in each panel has failed. In the portrait
of the man, the detached section of the panel was simply pinned at a later
date with nails of later construction (Figs. 4, 5). At some time before
the 1730s, in common with a large number of paintings in the collection,
the panels were painted on the reverse with a red, probably ochreous,
paint.
6
This paint layer runs into the split and over the back of the frame,
suggesting that the joint had already failed by 1730, as a result of wide
uctuations of RH possibly caused by relocation of the painting; perhaps
it was removed to a less well buered area of the house, such as the attic
or servants quarters, when the style of the portraits became unfashion-
able or the signicance of the sitters was forgotten. The depth of the
rabbet of the frame is only about 2 mm greater than the average 6 mm
thickness of the panels, indicating that no space for movement was
allowed for by the frame maker.
A seventeenth-century view of a church interior after Neefs,
from Grimesthorpe Castle in Lincolnshire, has horizontally aligned planks
(Fig. 6). The panel has a history of structural failure, as the uppermost
join has opened and has been reglued while misaligned. This initial failure
436 Mc Cl ure
Fi gure 4, bel ow
British school, Portrait of Edward Altham, 1617.
Oil on oak panel, 78.8 63.5 cm. National
Trust, Kingston Lacy House, Dorset, England.
The condition before treatment is shown; the
frame is original.
Fi gure 5, bel ow ri ght
British school, Portrait of Edward Altham.
Reverse before removal from the frame.
was followed by the present failure of the lower join; on both occasions
these failures were caused by the rigid xing of the panel into the frame.
Regular nicks on the sides of the reverse of the panel, cut for xings, can
be seen (Fig. 7).
An unthinned panel of approximately 1540, Portrait of a Man with
a Watch (Science Museum, London), attributed to the Florentine painter
Maso di san Friano, has regular V-shaped nicks along the top and bottom
of the reverse of the panel where nails have secured it to the frame. There
is no evidence of similar xings along the sides. The panel has developed a
convex warp, greatest in the center of the panel, between the dovetailed
battens, set into channels (Fig. 8).
437 Tnt Fraxi o or Woont Patis
Fi gure 6
Dutch school, Church Interior, midseventeenth
century. Oil on oak panel, 89.2 120.8 cm.
Grimesthorpe Castle, Lincolnshire, England.
Fi gure 7
Dutch school, Church Interior. Reverse before
removal from the frame.
The correction of the tendency of panels to develop convex warps
when exposed to low ambient levels of RH is probably one reason for the
increasing popularity of cradling from the mideighteenth century into
the rst half of the twentieth century. A panel would be thinned by the
introduction of moisture and become very responsive to attening. A
cradle would then be attached, to restrain the panel in plane. In this way,
the visual disturbance caused by a gap between the frame rabbet and the
picture surface could be corrected. By the midnineteenth century, panels
were routinely thinned and cradled, and even as late as the 1960s the desir-
ability of a at picture plane was cited as a reason for major intervention.
7
Today it is generally accepted that panel paintings should be allowed to
assume a natural warp at a given RH and that the frame should be
adjusted to suit such movement.
The principles behind early-twentieth-century solutions to the
problems of framing panels hardly dier from those we recognize today.
In 1936 in the National Gallery of Scotland, the wings of the Trinity
Chapel Altarpiece relied on the provision of a microclimate enclosure,
the exterior of which served as the frame (Cursiter 1936:10916). In 1940
the International Oce of Museums recommended, among other urgent
concerns, the use of steel springs in framing panels to allow movement and
added the proviso that they should be removed for transport (International
Oce of Museums 1940:80, 81, n. 58, 59). In 1955 George Stout, as part of
the survey of panel treatments instigated by the International Council
of Museums Commission for the Care of Paintings, illustrated examples
of frames causing splits in panels and of panels causing the breaking of
frames. Several systems for tting panels were illustrated, including a sys-
tem for supporting a panel with unglued cracks.
8
In 1965 Straub recom-
mended the use of exible strips of sprung steel to allow warping. This
apparatus could be combined with a backing to provide protection against
shock and to act a buer against changing climatic conditions (Straub 1965).
Similarly, in 1978 Goetghebeur recognized the use of the picture frame to
support panels and suggested the use of sprung steel strips to allow move-
ment (Goetghebeur 1978). In 1982 Ranacher largely repeated Straubs
recommendations (Ranacher 1982:147); the same year Vhringer described
a framing system for a sixteenth-century panel which supported the panel
in the center and allowed movement at the edges by means of a leaf spring
held in place by a U-section metal bracket and adjusted by a threaded bolt
(Vhringer 1982:g. 9). In 1988 Dunkerton and coworkers described the
widening of the groove of a later double-sided frame housing the wings of
an altarpiece by Martin van Heemskerck (Dunkerton, Burnstock, and Smith
1988:20). Low-density foam was tted in grooves on both sides of the panel
to allow some movement. Hermesdorf, in 1989, described a system of sus-
pending the panel on aluminium strips attached to the frame. Wooden
buttons, normally reinforcing reglued joins and splits, had slots cut in them
that tted over the aluminum sections (Hermesdorf 1989:26769). The use
of roller bearings attached to the base of a support for a large panel with
horizontally aligned planks and running on tracks on the bottom rabbet
section of the frame signicantly reduces static friction between the panel
base and frame, allowing the panel to move in response to changes in RH
(see Bobak, A Flexible Unattached Auxiliary Support, and Marchant,
Development of a Flexible Attached Auxiliary Support, herein).
These methods reect the current belief that panel paintings
should be allowed to move, perhaps within limits, to adopt greater or
438 Mc Cl ure
Fi gure 8
Maso di san Friano (attrib.), Portrait of a Man
with a Watch, midsixteenth century. Reverse.
Oil on poplar panel, 117 92 cm. Science
Museum, London.
smaller curves in response to changes in ambient RH. Observation of par-
ticular panels under varying levels of RH can reveal a surprising degree of
warping over a short period of time. For example, in 1987 the Hamilton
Kerr Institute, in Cambridge, England, treated an early-sixteenth-century
poplar panel, The Adoration of the Shepherds, attributed to the Master of
Santa Lucia sul Prato. The panel, measuring 169 162.5 cm, had been
thinned to 1 cm and was heavily cradled. After the cradle was removed
and the splits were reglued, the panel developed a convex warp of 3 cm
across its width at an RH of about 55%. A reduction or increase of 10%
produced an increase or reduction of the warp by 1 cm in just two hours.
Rigid xing of the panel in its frame would have inevitably produced rup-
ture of the wood or of the glued joins.
9
Stout, in describing a formula to
calculate the force required to constrain and atten a warped panel tem-
porarily straightened by moisture, gave a formula to calculate the force
required to rupture the panel; that formula could then be used to calculate
a safety margin in framing (Stout 1955:15859). In practice, however,
evaluation of the force required for the panel to deect the frame xings
seems only recently to have been assessed. In 1991 Mecklenburg and
Tumosa produced computer models of cracked and uncracked oak panels
rigidly xed into their frames and assessed their resistance to splitting. An
uncracked oak panel measuring 76 102 1.27 cm thick could be split by
uctuations of RH between 70% and 10%. When a cracked panel of simi-
lar dimensions is subjected to strain, much less force is required to extend
the cracks (Mecklenburg and Tumosa 1991:187.). Thinned poplar panels,
often weakened to a far greater extent than oak by boring insects, are
likely to split under much lighter loads.
The author has not found any assessment of the force exerted
either by exible spring xings, commonly used to secure panel paintings,
or by malleable brass xings, which might be expected to distort under
loading, thereby preventing undue stress to the panel or elastic foam of a
known density (Plastazote and Evazote of a density of 50 kg m
3
are
commonly used).
10
However, a simple experiment demonstrates that the
force required to deect a particular xing is much greater than might be
expected. Two commonly used sprung-steel xings and four brass xings
of dierent dimensions were screwed to a length of wood. Holes (or in
the case of one spring xing, a hook) were provided to attach a spring
balance. The force required to raise a xing by 1 cm was observed. The
length of each xing, which aects the moment of the force generated,
was not assessed. The spring xings required a loading of 0.81.4 kg to
deect them 10 mm from an unstressed position. A force of 2.8 kg was
required to move the three smaller malleable brass strips 10 mm; a force of
1.8 kg was required to move the largest brass xing, a result that reected
the increasing moment as the length increased. Foam blocks of varying
density require an often-underestimated force to compress the foam to
accommodate a warp. For example, three foam blocks, 3 cm cubes cut
from Evazote of 50 kg m
3
in density and set in line at 10 cm centers
under a strip of wood, required a weight of 7.3 kg to compress the blocks
by 5 mm. Two identical blocks at 10 cm centers under the same wooden
strip required a weight of 5.5 kg to compress them by 5 mm. A single
foam block under the wooden strip required a weight of 2.7 kg to com-
press it by 5 mm. The force exerted by the metal xing devices described
above, when the panel moves against them, could in many cases come
close to or exceed the rupture strength of the wood, especially when the
439 Tnt Fraxi o or Woont Patis
wood is weakened by splits or degraded by woodworm damage, and when
glued joints are embrittled by age. It is doubtful that any calculation could
be devised to give a value for the elasticity of the wooden panel at right
angles to the grainthere being so many variables and features peculiar to
each panel. It seems that framing systems that exert minimal restraining
force at the edges and parallel to the wood grain are least likely to cause
damage. A framing system to achieve such a goal would have to be
designed on an individual basis, with the construction and inherent stresses
of each panel, as well as its display and travel requirements, taken into
account. It is an important consideration that systems that hold the panel
while allowing maximum movement often provide insucient support
when the panel is moved.
Three case studies illustrate techniques employed at the Hamilton
Kerr Institute. The rst, the framing of a small panel of three vertical
planks, the Portrait of Elizabeth Altham, has been mentioned above. The
second is the framing of the large altarpiece Noli Me Tangere by Anton
Raphael Mengs, with horizontally aligned planks; the frame of that work
required considerable strengthening. The third is an altarpiece, attributed
to Pietro Gerini, which had been reframed in the nineteenth century.
Even though the frame was causing splitting in the three panels, the client
wished the altarpiece to remain unchanged in appearance after treatment.
The 1617 panel, one of a pair of seventeenth-century portraits, was in its
original frame of oak painted black and partly gilded. The joints of the
panel had opened. After gluing, the panel was observed to assess the maxi-
mum curvature it would develop. It became slightly convex at an RH of
50%. It was decided to increase the depth of the frame rabbet from 8 mm
to 18 mm. Strips of dimensionally stable spruce were cut and angled to the
outside edges of the frame to make the addition inconspicuous. Strips 25
mm in width were mitered, stained dark, and attached with screws to the
back of the frame. (The use of glue was rejected as less reversible.) The
addition is not visible when the frame is hanging, and only a slightly larger
gap between frame and wall is evident. The panel tted quite closely to
the rabbet, which did not require any addition of shaped sections to follow
the panels curvature. Rabbets can be adjusted to remove the gap between
the sides of the panel and the frame edge, which can be visually distract-
ing, especially where light can be seen between the picture frame and the
wall. However, a curved rabbet can itself restrict movement if the panel is
subjected to higher RH levels; in such a case the sides of the panel across
the grain will press against the outer edge of the rabbet and the nearest
top and bottom xing points, exerting pressure on weak areas and joins.
The panel must be able to assume a less convex prole; this is facilitated
either when space is left for movement at the edges or when the central
xings are designed to compress, allowing the panel to move away from
the frame rabbet in the center. Compressible curved additions to the rab-
bet could be made. However, any material that can be accurately shaped
and that presents a visually acceptable surface, such as Plastazote or
Evazote, is likely to be too rigid to conform to changes to the congura-
tion of the panel.
The panel was then tted in the frame. The central vertical plank
was set on a thin, 15 mm strip of hardwood. This raised the lower edges of
the panel away from the bottom edges of the frame, allowing free move-
Portrait of Elizabeth Altham
440 Mc Cl ure
ment. Small blocks of Evazote, with the bottom edges shaped to align
with the chamfered edge of the panel, were cut. A groove was cut in the
top of the blocks to position the brass xing strip and prevent the foam
block from accidentally falling out. The brass xing strip was bent at a
right angle at its end to retain the block. It was then screwed to the base of
the frame addition and curved to hook over the back edge of the addition
so that it could not be twisted out of position. These xings were placed
toward each edge of a central plank (Fig. 9).
The structural conservation of this large painting is described elsewhere
(see Brewer, Some Rejoining Methods, herein). Even after an auxiliary
support in the form of battens was applied to the reverse, it was felt that
the frame should supply further support. As the weight of the panel itself
is estimated to be about 100 kg and the planks are horizontally aligned, it
was decided to modify the system designed by Ray Marchant for a painting
by Alexander Kierincx and Roelant Savery (see Marchant, Development
of a Flexible Attached Auxiliary Support, herein). This system would
reduce the friction between the bottom edge of the panel and the frame
by attaching roller bearings to the panel, locating them in slots in the
depth of the frame rabbet. The existing frame originally had no rabbet,
and so later additions were removed, and a new back rail, 19 cm in depth,
was made to accommodate the panel and batten system and a movement
of about 2 cm in either direction. This back rail was glued and screwed to
the back of the frame for maximum strength and painted to match the
sides of the frame. The lower ends of the four battens supporting the pic-
ture were then tted with inserts, so that an L-shaped aluminum section
could be bolted on to support the weight of the panel when upright. A
pine spacer shaped to counter the angled edge of the panel due to the con-
vex warp was tted between the base of the panel and the aluminum sec-
tion to spread the weight of the panel evenly. The L-shaped aluminum
Noli Me Tangere by
Anton Raphael Mengs
441 Tnt Fraxi o or Woont Patis
Fi gure 9
British school, Portrait of Elizabeth Altham,
1617. Reverse. Oil on oak panel, 79 x 63.5 cm.
National Trust, Kingston Lacy House, Dorset,
England. Modications to the original frame
are shown.
section protruded in front of the panel by 5 mm to enable the block hold-
ing the roller bearings to be set in the line of the center of gravity of the
panel. The blocks of aluminum were bolted to a strip of 6 mm aluminum,
which was bolted in turn to the L-shaped aluminum section to increase
rigidity. The bearings run in the slots cut in the bottom rabbet of the
frame, which are lined with stainless steel strips to prevent the bearings
from denting the wood and locking the system. When upright, the panel
rests on the bottom edge of the frame and against shaped rabbets on
either side. The panel is held in the frame across the center by two hori-
zontal battens resting on the panel battens and attached to the frame by
bolts and compression springs to allow some movement in the event the
panel should assume a less convex prole (Figs. 10, 11a, b).
When received for treatment, the three elements of the triptycha central
single poplar panel of the Madonna and child, anked by panels of two
vertical planks with arcading attached abovewere held together by three
battens screwed to each panel in turn (Hamilton Kerr Institute 1984)
(Figs. 12, 13). At the front, the decorated base was screwed to the lower
edges of the panel. This rigid construction, probably made in the late
nineteenth century in France, as the provenance suggests, had caused the
opening of splits in the left and right panels and two splits in the central
panel. The two columns separating the wings from the center were con-
temporary with the rest of the frame.
The components of the altarpiece were dismantled. The battens
were unscrewed and the columns and pilasters gently prized o. A new
framework to hold the panels was constructed from pine, stained to be
unobtrusive, and then sealed with polyurethane varnish to reduce dimen-
sional change in the structure (Fig. 14). The frames base was adjusted
with balsa-wood spacers to hold the three panels at the correct alignment
(Fig. 15). Each outside panel was attached by the base of the support and
by a shaped brass strip xed to the back of the framework at the upper
rail. The central panel was similarly attached at the base and attached by
steel hooks to existing original xings in the panel above the pilasters. Two
vertical strips of wood were placed over the joins between the central
panel and the outer panels and were held in place by stainless steel bolts
Triptych attributed to
Pietro Gerini
442 Mc Cl ure
Fi gure 10
Anton Raphael Mengs, Noli Me Tangere, 1771.
Reverse. Oil on walnut panel, 291.5 178.5
cm. All Souls College, Oxford.
that passed through the gap between the three panels and that were bolted
to the framework with provision for adjustment. The three panels were
thus held and supported without any restriction to their movement.
To comply with the clients wishes, the nineteenth-century
columns and base were put back. The decorated base was attached to
the base of the framework, the bases of the columns at each end were
443 Tnt Fraxi o or Woont Patis
Panel (horizontal planks)
Batten
6-mm bolts into wooden inserts
Addition to frame rabbet
Aluminum L section
Wooden strip to square end of panel
Aluminum plate
Aluminum housing
Roller bearing
End stop
Stainless steel plate to
spread load
Back rail of frame
Frame
Bolt
Washer
Spring
Washer
Aluminum sleeve
covering bolt
Wooden insert
Back rail of frame
Frame
Felt tape
Shaped profile in wood
following curve of panel
Cross battens
Batten
Panel
Batten
Fi gure 11a, b
Anton Raphael Mengs, Noli Me Tangere.
Diagram of (a) roller bearing xings, and
(b) sprung framing battens.
a
b
lengthened, and the pilasters above the spring of the arches were rexed
(those on either side of the central panel to that panel only). The sides of
the framework were chamfered to echo the shape of the columns, and
new central columns were made to cover the strips supporting the panels
on either side of the central panel. The columns, slotted behind the capi-
tals and held in place at the bottom by a thin base plate screwed onto the
base of the framework, were prepared and gilded to match the rest of
the framing elements. The reframed altarpiece diered only in minute
detail from its previous appearanceyet the panels were unrestricted,
and the frame could be easily dismantled and reassembled (Fig. 16).
It is hoped that these examples demonstrate a valid and exible
approach to the framing of panels. The solutions devised here are not pre-
sented as models to be copied but, rather, as proposed methods that can
be adapted and improved.
444 Mc Cl ure
Fi gure 12, above
Pietro Gerini (attrib.), Triptych. Acqueous
medium on poplar panel, 165 180 cm.
Private collection. The condition before treat-
ment is shown.
Fi gure 13, above ri ght
Pietro Gerini (attrib.), Triptych. Reverse
before treatment.
Fi gure 14, bel ow
Pietro Gerini (attrib.), Triptych. Reverse
after treatment.
Fi gure 15, bel ow ri ght
Pietro Gerini (attrib.), Triptych. Diagram of
the framework.
The author is grateful to Kathryn Hebb for her help with this article, as
well as to Ray Marchant for his help and advice.
1 For a well-illustrated history of frame styles, see Grimm 1978. An important study of Dutch
frames in the seventeenth century is van Thiel and Kops 1984.
2 National Gallery, London, cats. 56978. A full account of the technique and construction of
the altarpiece can be found in Bomford et al. 1989, especially 156.
3 For a succinct account of the construction of early Italian altarpieces, see the introduction by
Bisacca and Kanter in Newbery, Bisacca, and Kanter 1990:1130. See also Cmmerer-George
(1966) for illustrations of altarpiece construction.
4 See Lacey 1970:6580, for a study of the environment in Kings College Chapel, Cambridge,
before the installation of the large panel of The Adoration of the Magi by Rubens. The consider-
able buering eect of the stone is assessed, as are the eects of low-level winter heating,
which contributed to an annual uctuation of mean levels of RH between 55% and 70%, with
very slow rates of change.
5 For a detailed account of Flemish altarpiece construction, see Verougstraete-Marcq and Van
Schoute 1989:78, where the problems encountered with the original design of the wings of
the Ghent Altarpiece are discussed.
6 A list of paintings mended and cleaned by George Dowdney in 1731 survives. The linings of
several of the Lely portraits were eighteenth century and had been coated with red paint on
the reverse, presumably as a moisture barrier. See Laing 1993:10731, especially n. 18.
7 For example, Helmut Ruhemanns comment on the semi-transfer of the Nativity by Piero
della Francesca, which did not leave the picture quite at nor absolutely stable (Ruhemann
1968:161, n. 2).
8 Stout 1955:gs. 2224. For tting panels in frames, see gs. 50, 52, and 53.
9 See Hamilton Kerr Institute 1987, where the work is attributed to the school of Ghirlandaio.
The painting is in a private collection.
10 Plastazote is a low-density, cross-linked, closed-cell polyethylene foam. Evazote is a low-
density, cross-linked, closed-cell ethylene vinyl acetate foam.
Notes
Acknowledgments
445 Tnt Fraxi o or Woont Patis
Fi gure 16
Pietro Gerini (attrib.), Triptych. The condition
after treatment is shown.
Brass xing strips of various sizes, J. Shiner and Sons, 8 Windmill Street, London
W1P 1HF, England.
Plastazote and Evazote, BXL Plastics, ERP Division, Mitcham Road, Croydon, Surrey
CR9 3AL, England.
Roller bearings, Always Engineering, Warner Street, Birmingham, B12 0JG, England.
Self-adhesive acrylic felt tape (for lining rabbets of frames), George B. Tewes Co., Western Felt
and Fiber, 323 South Date Avenue, Alhambra, CA 91803.
Bomford, David, Jill Dunkerton, Gordon Dillian, and Roy Ashok
1989 Art in the Making: Italian Painting before 1400. London: National Gallery.
Cmmerer-George, Monika
1966 Die Rahmung der Toskanischen Altarbilder im Trecento. Strasbourg: P. H. Heitz.
Cursiter, Stanley
1936 Notes: Control of air in cases and frames. Technical Studies in the Field of Fine Arts
5(October):10916.
Dunkerton, Jill, Aviva Burnstock, and Alastair Smith
1988 Two wings of an altarpiece by Martin Van Heemskerck. National Gallery Technical
Bulletin 12:1635.
Goetghebeur, Nicole
1978 Treatment of panels at the Institut Royal de Patrimoine Artistique. In Conservation of
Wood in Painting and the Decorative Arts: Preprints of the Contributions to the Oxford
Congress, 1723 September 1978, ed. N. S. Brommelle, Anne Moncrie, and Perry
Smith, 16568. London: International Institute for the Conservation of Historic and
Artistic Works.
Grimm, Claus
1978 Late Bilderrahmen. Munich: D. W. Callwey.
Hamilton Kerr Institute
1984 Triptych, attrib. Pietro Gerini. Conservation Report 783, Hamilton Kerr Institute,
Cambridge, England.
1987 The Adoration of the Shepherds, School of Ghirlandaio. Conservation Report 992,
Hamilton Kerr Institute, Cambridge, England.
Hermesdorf, Peter
1989 Konservierung und Montage bemalter Holztafeln. Restauro 95(4):26769.
International Oce of Museums
1940 The Manual on the Conservation of Paintings. Paris: International Oce of Museums.
Lacey, Ralph
1970 A note on the climate of a medieval chapel. Studies in Conservation 15:6580.
Laing, Alastair
1993 Sir Peter Lely and Sir Ralph Bankes. In Art and Patronage in the Caroline Court, ed. David
Howarth, 10731. Cambridge: Cambridge University Press.
McClure, Ian, and Renate Woudhuysen
1994 The Oxburgh Chapel Altarpiece: Examination and conservation. Apollo 139(387):2023.
References
Materials and Suppliers
446 Mc Cl ure
Mecklenburg, Marion, and Charles Tumosa
1991 Mechanical behaviour of paintings subjected to changes in temperature and relative
humidity. In Art in Transit: Studies in the Transport of Paintings, ed. Marion
Mecklenburg, 173216. Washington, D.C.: National Gallery of Art.
Newbery, Timothy, George Bisacca, and Laurence Kanter
1990 Italian Renaissance Frames. New York: Metropolitan Museum of Art.
Ranacher, Maria
1982 Gemldebefestigung im Rahmen. Restauratorenbltter 6:147.
Ruhemann, Helmut
1968 The Cleaning of Paintings. London: Faber and Faber.
Smith, Alastair, Anthony Reeve, Christine Powell, and Aviva Burnstock
1989 An altarpiece and its frame. National Gallery Technical Bulletin 13:2843.
Stout, George, ed.
1955 The care of wood panels. Museum 8:13994.
Straub, Rolf
1965 Einrahmung (Section S), part 1 (Holztafeln). In Konzervierung und Denkmalpflege.
Zurich: Institut fr Kunstwissenschaft.
van Theil, P. J. J., and C. J. de Bruyn Kops, eds.
1984 Prijst de Lijst. Amsterdam: Rijksmuseum.
Verougstraete-Marcq, Hlne, and Roger Van Schoute
1989 Cadres et supports dans la peinture flamande aux 15e et 16e sicles. Heure-le-Romain,
Belgium: H. Verougstraete-Marcq.
Vhringer, Brigitte
1982 Praktische Mglichkeiten der Restaurierung von Holztafelbildern am Beispiel der
Gebots-Tafel aus der Kreuzkirche in Dresden. In Beitrge zur Erhaltung von Kunstwerken,
ed. Ingo Timon, Hans-Joachim Granan, Angela Mlles, and Christine Heidenreich,
95103. Berlin: Verband Bildener Knstler.
447 Tnt Fraxi o or Woont Patis
448
A
sutsrari ai rrororri o of paintings treated at the Hamilton
Kerr Institute (HKI) in Cambridge, England, have been on wooden
supports. This article uses examples to show the underlying causes
and mechanisms that determine treatment decisions in practice.
Scales of damage and treatment constraints
Large panels have sucient weight and size in the cross-grain dimension
1
so that a number of considerations arise that are generally less signicant
in smaller panels. Greater damages are foundbreaks of greater number
and length and larger areas of biological deteriorationwith correspond-
ing treatment implications. Liters of (usually toxic) consolidant may be
needed for a large volume of insect-damaged wood, requiring large-scale
application methods and large-capacity fume extraction. Thus, treatment
methods are scaled accordingly and should be made as ecient as possible,
while, of course, being subject to conservation demands.
Structural stabilization concerns for large panels must be balanced
with restrictions in time, cost, and methodology. The greater logistics gen-
erally make treatment more dicult, demand more time and appropriate
methods, and therefore increase total costs. Satisfactory results may
require a complex treatment and some ingenuity.
Environmental considerations and wood movement
Environmental conditions are constantly changing, however slightly, so that
panels of wood are constantly moving in response to changing moisture
content (MC).
2
Depending on the panel structure, such wood movement may
be relatively small or slow, and therefore not easily perceived. Generally,
larger panels change MC more slowly, although movement may still be rel-
atively fast, especially for thinner panels. Even when housed with the best
environmental controls, panel paintings are unlikely to reach a stable equi-
librium moisture content (EMC) with level moisture gradients and cessa-
tion of movement.
Relative humidity (RH) should be as stable as possible during
treatment. Total lateral movement (across the wood grain and in the plane
of the panel)
3
and warp movement (perpendicular to the panel plane)
vary directly with the panels dimension across the grain.
4
The location
General Considerations
Practical Aspects of the Structural Conservation
of Large Panel Paintings
Al Brewer

of warp-prone areas is also a factor. For example, planks cut tangentially
are more prone to warp movement. If a warp-prone area is located
toward the middle of a panel, the movement will be transmitted to the
panels (longitudinal-grain) edges so that the overall deection may be
somewhat greater than that of the central plank (Fig. 1).
5
Since this is an
angular relationship, deection of the panel edges may be almost instan-
taneous, especially for larger, thinner panels that are more exible and
therefore more responsive.
6
Treatment RH should be similar to that of the panels normal or
destined location (Fig. 2). If not, after the panel is relocated there will be
further movement opposing any restraints imposed by rejoining, reinforce-
ment, or framing. Eective treatment should lessen potential stresses in
the painting structure as much as possible.
Proportional increase in total wood movement has other implica-
tions for panels of larger cross-grain dimensions. The development of
end-grain splits or checks is well-known in the drying of commercial oak
timber, especially larger sections. This is partly due to much higher mois-
ture permeability through end-grain, where oaks large-diameter vessels
play a part, than through side-grain surfaces.
7
A similar phenomenon
seems evident with respect to wood movement in oak panel paintings
where cyclic compression sets and tensions provoke end-grain fractures
(Desch 1956:9395) and disjoins. These eects are proportionally greater in
wider planks. Like oak, walnut has relatively high density and large vessels.
Figure 3 shows a joint between wide walnut planks that had parted several
times, developing an ever-increasing gap, evident from the stratigraphy of
three or four putty layers.
8
Structure of larger panels
Structure determines many aspects of conservation. Tree species that
grow larger and yield larger planks have usually been used in large panels.
White poplar (Populus alba L.), oak (Quercus spp.), and Scots pine (Pinus
sylvestris L.) are examples. Large panels are sometimes made from rela-
tively small planks, as in some of Rubenss larger landscapes (Brown,
Reeve, and Wyld 1982).
449 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
Dimension 1
Dimension 2
Deflection 1 Deflection 2
Fi gure 1, above
A panel, consisting of three planks, viewed
from slightly above one end of the planks. A
greater cross-grain dimension would mag-
nify the deection of the middle plank in the
diagram. This is shown as the dierence
between deection 1, for a smaller panel of
dimension 1, and deection 2, for a larger
panel of dimension 2.
Fi gure 2, ri ght
A tent enclosure for treatment of a large
panel, built to sustain RH and temperature at
approximately the same levels as in the panels
normal environment.
Fi gure 3
A gap in a joint toward one end of the adjoin-
ing planks shown after several strata of over-
lying nonoriginal layers and putty were
removed. Since disjoining, the plank edges
had developed a step that had been subse-
quently abraded to the same level. The length
of joint shown is about 300 mm.
Some unusual woods may be found. Raphaels Transguration,
9
painted on cherry wood (Prunus arium L.), is a very large panel (Mancinelli
1990:150). Italian panels are usually associated with white poplar (Bomford
et al. 1989:11), linden (Tilia spp.) (Klein and Bauch 1990), and perhaps wil-
low (Salix spp.),
10
so a very large panel of cherry is unusual.
Figure 4ac shows an Italian painting on white poplar consisting
of six vertical planks cut to a round top. Though it seemed slightly heavy,
the wood density is within poplars rather broad range ( Jobling 1990:66).
This panel would originally have been about 4050 mm thick. Poplar has
been justly preferred for large panels because of its high strength-to-
weight ratio, ease of tooling and preparation, and moderate movement
(Lincoln 1986:221).
In rare examples, one or two large planks may suce for a
support. Figure 5a, b shows a painting by J. M. W. Turner on a single
mahogany plank. Its sister painting is similarly constructed.
11
Both panel
paintings remain in extremely good condition, which is not unusual for
sound mahogany panels
12
in a near-original statethat is, unaltered by
thinning, cutting down, and so forth. Neither is restrained by an auxiliary
support. If excessive, such restraint can be detrimental, particularly for
larger panels. Movement of mahogany is small (Lincoln 1986:159), an
advantage for preservation. With radially sawn planks and sound prepara-
tion, such panel paintings tend to be durable.
In portrait format, vertical planks are a more structurally sound
arrangement than horizontal planks. Generally, rectangular panels have
450 Bre we r
Fi gure 4ac (4c on next page)
Marco Palmezzano, The Mystic Marriage of
Saint Catherine, 1537. Oil and egg tempera on
poplar panel (visual identication), 2560
1805 20 mm thick. Property of the Marquess
of Northampton. Before treatment, front with
frame rubs along the curved edge (a); back (b);
and a diagram of reinforcements and splits
viewed from the back (c). Some splits were ini-
tiated or aggravated by screw tips protruding
from the framework and into the panel back
in the lower right corner, and especially by a
cross-grain insert at the middle of the bottom
edge, where the splits zigzag abruptly.
a b
Panel
Earliest
Planks Splits
Dowelled
lattice
Panel
Earliest
Planks Splits
Dowelled
lattice
planks joined parallel to the longer edge, decreasing the work necessary
for assembly.
13
The Visitation
14
by Tommaso Manzuoli (Keith 1994) and the
Transguration by Raphael are both large by any standard. Most of the ver-
tical planks of the Manzuoli are extended and consist of two planks joined
end to end. In contrast, the planks of the Raphael are remarkably long,
and they are not extended.
Large panels in portrait format with horizontally disposed planks
are more prone to structural problems. Greater and more concentrated
weight and greater warp movement provoke damage. For example, a
large panel by Vittore Carpaccio,
15
though still of approximately original
451 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
c
thickness, shows considerable joint damage. Despite these observations,
Figure 6ac shows a painting by Rubens in Kings College Chapel,
Cambridge, that has planks joined horizontally in portrait format. It is
in very good condition. The oak planks are of original thickness; this has
preserved the original joint surfaces and minimized wood movement so
that the panel has remained at overall.
In landscape format, horizontal planks perform better. Joint
strength is increased by greater surface area, and the panels center of
gravity is lower.
Handling of larger panels
In combination with greater weight, the ratio of cross-grain dimension to
thickness is usually large enough to give larger panels a greater tendency to
bend when handled or inadequately supported. In other words, for a con-
stant thickness, panels that are larger across the grain become more prone
to bending and subsequent damage. All panels bend when handled, though
bending is not always perceived. Sound panels may withstand considerable
bending stress.
Restraint considerations are more acute for larger panels, because
of greater total movement, greater potential leverages, and weight eects.
Restraint of moisture-dependent movement, such as that imposed by restric-
tive framing or reinforcement structures, can increase stresses. A statically
restrained panel may be under considerable stress. Momentumalso a
greater factor when heavier panels are handled or transportedshould be
452 Bre we r
Fi gure 5a, b
J. M. W. Turner, Story of Apollo and Daphne,
exhibited 1837. Oil on mahogany panel, 1100
1990 10 (bevel) to 20 mm (middle) thick.
Tate Gallery, London. Front (a) and back (b)
of a large panel consisting of a single
mahogany plank.
a
b
considered in relation to potential stresses, whether a panel is restrained
statically or allowed to move more freely.
Common sense should dictate precautions in handling. People
with experience in panel structural work tend to rely on a sense of feel
when handling a panel or judging its strength. This sense is probably a
combination of experience, touch, and a keen attention to and awareness
of the physical nature of the object. Inexperienced or careless handlers
may be overcondent or, conversely, too cautious.
Therefore, when larger panels are moved, it is better to have at least
one person present who is experienced in handling such objects. Two or
more people are needed to move larger panels any distance. Coordination
is important, because it is dicult to sense and maintain a constant share of
the weight. A person on one corner of the panel may allow that side to
droop, thereby causing a dangerous bend or twist.
Most wood is much weaker in the cross-grain direction. Strength
in axial tension is up to fty times greater than in tension perpendicular to
the grain (Tsoumis 1991:162). If a panel is moved toward the horizontal,
its weight should not be supported only at the side-grain edges
16
or only at
the middle of the end-grain edges. In the rst situation, the weight causes
sagging of the middle. In the second, the sides sag. In both cases, the panel
must be structurally sound (i.e., no major defects) to withstand cross-grain
bending safely. To support and balance weight better in the stronger axial-
grain direction, greater support should be given along the end-grain edges,
primarily at about one-quarter to one-third the distance from each corner.
Further support may help decrease stress.
Panels in a vertical position are usually handled by the sides,
which are usually the longitudinal-grain edges. This happens with larger
panels because people must usually stand at the sides to lift. If the panel is
tipped to lie horizontally, then the grip can be shifted to a better position
at the end-grain edges to avoid the bending stresses discussed above.
If it is necessary to move a large panel from one edge to another,
it is safer to lower it to a horizontal position and then to raise it again onto
453 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
a
b c
Fi gure 6ac
Peter Paul Rubens, The Adoration of the Magi,
seventeenth century. Oil on oak panel, 3280
2465 mm; original thickness unknown. Kings
College, Cambridge. Front (a); one of three
vertical battens of the original auxiliary sup-
port (b); and an original iron cleat (c). Note
the channel cut in the batten (c) to allow
panel movement across the grain (vertically);
the upper horizontal line is a joint. The panel
has been maintained in a large interior space
under relatively stable conditions. Disjoins
along the side edges extend relatively short
distances inward and stop near or at the first
vertical battens. They had been treated with
inserts or bridges. The battens, which are of
relatively small rectangular section, do not
appear to have caused excessive restraint,
since the nailed cleats remain relatively
unaffected.
the desired edge rather than to cartwheel the panel from one edge to
another. When leaning a large panel away from a vertical position, it is
also important to stop the bottom edge from sliding out. Weight directed
at an angle to the oor can cause uncontrollable slides.
The greater mass of larger panels creates greater inertia, so rapid
movements increase the likelihood of bends and twists. Twists or torsions
cause stresses from many angles and are probably the most dangerous pos-
sibility (Gordon 1978:chap. 12). In describing the twisting of plates in rela-
tion to thin wooden panels, Bodig and Jayne have noted that a body in
pure torsion is in a state of pure shear stress that is concentrated at the
upper and lower surfaces (Bodig and Jayne 1982:165). Essentially, when a
plate sustains a twist, one pair of diagonally opposite corners are forced
closer together, while the other two corners are forced further apart.
Boxlike structures or diagonals resist these distortions more eciently,
and they have been employed in some original reinforcements (Marette
1961:pls. 22, 23; Castelli and Ciatti 1989:14243).
Twists can occur even with seemingly robust auxiliary supports,
such as thick battens. The additional weight of the auxiliary support can
increase the danger, as it is usually supported to some degree by the panel
itself. A thicker panel resists bending and torsion better.
It is usually better to carry large panels with the grain held verti-
cally. They should rest on an end-grain edge and lean slightly away from
the painted side. During handling, great care should be taken if panels
are laid horizontally or on a longitudinal-grain edge. The momentum
of movement is transferred more dangerously to a horizontal panel,
while buckling may occur in the latter case. Over longer distances, well-
supported trolleys alleviate stresses on panels and bearers (Fig. 7). It is
important to have the route clear and to have the panels destination pre-
pared for both breadth and height.
Temporary auxiliary supports
It may be useful to build a temporary auxiliary support
17
if a large, weak-
ened panel is to be moved frequently or treated extensively. Designs can be
tailored accordingly.
454 Bre we r
Fi gure 7
A large panel being transported on a
custom-built trolley, attended by qualied
personnel (1987).
A Flemish oak panel of original thickness had two nonoriginal
metal battens screwed to the back (Fig. 8ac). Not surprisingly, the panel
developed splits and disjoins. A chalky, weak ground, combined with
restrained wood movement, had caused tented aking and losses. The
battens were removed to prevent further damage while the painting
awaited treatment.
To remove the battens, the panel was laid horizontally for better
control. There was concern that release of the battens might cause a sud-
den warp movement and precipitate further aking. A temporary frame-
work was built to allow the panel to assume an unrestrained shape, as well
as to provide support, improved access, and secure handling (Fig. 8d).
Wood may be preferred for such temporary supports, since a basic
framework can be built quickly and easily. An adjustable, reusable, and
therefore economical alternative was built from wood and right-angled-
section metal girders, slotted for bolted assembly.
18
A smaller honeycomb-
core panel was bolted to the middle of the framework to preserve some
access from below and to decrease twist.
Adjustable levelers, made from machine bolts threaded into brass
plates, were attached to the framework crossbars in a regular pattern
(Fig. 8e). The levelers were turned against exible wooden battens that
conformed to the panel back. As the metal battens were removed, the lev-
elers were periodically readjusted to maintain contact as the panel changed
shape. Fortunately, little movement occurred in this case, but the screws
were readjusted periodically as the panel equilibrated.
This type of metal girder can be used for several purposes, such as
the trolley shown in Figure 7, which was later used as a trolley easel to
support a large panel for treatment. The pair of rubber wheels at one end
swiveled. A central pair was xed to roll parallel to the longer trolley axis
to allow easy maneuvering in any direction.
Mobility of such temporary supports is useful, especially in a busy
studio where large paintings must be moved often to allow photography,
passage of other large paintings, and so forth. For stationary support,
either the wheels were blocked, or the base was elevated slightly onto
wooden battens or bricks. More rigidity could be had by doubling the
girders or by adding more structure.
The structure and treatment of two large panels will be compared and
contrasted because they show an instructive range of dierences in period,
place of origin, materials, construction, changes over time, deterioration,
and conservation interventions. Their similarities show much about the
structural behavior of large panels. An attempt has been made to relate
the need for treatment, and some available treatment options, to the
causes of deterioration. Some points specic to each case are included to
emphasize the individuality that bears on treatment decisions. Though
neither panel is typical, their mechanisms of change are similar to other
cases. The paintings are referred to by the artists names.
Figure 9ac shows a painting by Anton Raphael Mengs (172879)
on walnut ( Juglans regia L.) that was completed in 1771 for the chapel of
All Souls College, Oxford.
19
The use of wood of a relatively high density is
slightly puzzling for such a large panel and, moreover, one that would have
had to be transported from southern Europe.
20
Though the painting is now
about half its original thickness, its original weight may be estimated to
Two Examples of
Large Panels
455 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
456 Bre we r
Fi gure 8ae
Ambrosius Francken I, The Judgment of
Zaleucus, late sixteenth century. Oil on
oak panel, 1795 2165 20 mm thick.
Fitzwilliam Museum, Cambridge (inv. 781);
HKI treatment no. 137. A large panel of origi-
nal thickness, damaged by attached (non-
original) metal battens, which restrained
wood movement: (a) front, faced, before treat-
ment; (b) back before treatment; (c) detail of
back, showing metal batten and one split;
(d) temporary reinforcement of metal frame
and honeycomb-core panel, to decrease twist;
(e) in one corner, an adjustable leveler bears
on a exible wooden batten, against which
the panel was laid.
a
b
c
d
e
457 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
Fi gure 9ad
Anton Raphael Mengs, Noli Me Tangere, 1771.
Oil on walnut panel, 2915 1785 20 mm
thick. All Souls College, Oxford. A large panel
treated while lying on a side edge on a mobile
temporary support: (a) front, before treatment,
showing disjoins; (b) the front upper right of
the panel, seen in top-raking light; (c) back
during treatment, showing the last planks of
the previous balsa laminate reinforcement
(bottom) and the panel surface, damaged by
thinning. The exposed mortises, the two
remaining original tenons, and areas of insect
damage and plaster ller can also be seen.
Note the ratcheted polyester straps used to
apply pressure during rejoining, and the
vertical angle-sectioned beam used with
veneer hold-down clamps to apply alignment
pressure. A diagram of the panel construction
(d) shows the tapered planks joined in reverse
orientation and the irregular gaps (exagger-
ated) that developed after the panel disjoined.
a
b c
1785
2915
44
343
281
284
461
398
376
30
1785
842
39
16
339
451
386
598
258
211
354
38
224
Plaster fill
Insect-damage
boundary
Original
tenon
Original
mortise
Joint gap
(exaggerated)
Joint
Holes
(All dimensions mm)
221
476
d
458 Bre we r
have been about 140 kg.
21
Painted in portrait format, the substantial
remaining weight and reduced thickness have had serious consequences.
The most recent conservation treatment, carried out in the 1960s,
included thinning and reinforcement with a balsa laminate similar to that
described by Lucas (1963). Subsequently, metal strips were added around
all edges. Many of the panels original joints later parted, presenting a
precarious structure and dismembering the image both literally and
guratively. Structural damage made reappraisal of the paintings con-
dition necessary as well, despite its recent restoration.
In contrast, a painting of 1537 by Marco Palmezzano (ca. 1458
1539) (Fig. 4ac) with a lower-density poplar construction (briey
described above) arrived with the image greatly obscured by darkened
varnish layers and surface dirt.
22
Weakened in areas by insect damage,
this panel had also been thinned to about half its original thickness,
which weakened it further. A lattice of wood had then been glued to
it, probably before it left Italy.
23
Undulations and compression damages
attested to poplars high capacity for bending and distortion under
mechanical stress. Fortunately, the painting exhibited sound technique,
a great advantage for structural conservation.
24
Both panels were assembled with casein glue. The joints of the
Palmezzano had remained intact under stress, while the weaker, brous
wood had parted into disconnected intermittent splits. Under stress, the
stronger and more rigid walnut of the Mengs had remained relatively
intact, while the joints had parted in the glue layer. These dierences in
fracture characteristics were due in part to the varying restraints imposed
by the auxiliary supports. The wood of both paintings had fractured pref-
erentially in insect-damaged areas.
More than thirty splits had developed throughout the
Palmezzano, mainly from movement-restricting battens glued to the back.
Some splits were older, with putties and aged varnish in the gaps, while
others were obviously recent, with freshly exposed and fractured ground.
What factors led to deterioration of these panel paintings? Both
panels may be examined more closely to understand the eects of struc-
ture, age, and past treatments on their condition.
Supports
The Mengs consists of six broad walnut planks arranged horizontally with
respect to the image and joined in reverse orientation (Fig. 9d).
25
The bot-
tom consists of two additional pieces: a narrower plank at the extreme
edge, joined to a narrow, wedge-shaped strip. The wedge was used to
square the bottom edge in relation to the taper of the lowest broad plank.
At the extreme top edge, there is a similar narrow plank but no wedge.
Evidence shows that the panel was originally about 40 mm thick,
twice its current thickness. The mortises and loose tenons had been
uncovered by modern tools during the most recent thinning. The mortises
had been chiseled into the joint faces to within 8 mm of the front of the
panel
26
so that if originally centered, they would have left the same thick-
ness of 8 mm at the back. The tenons
27
were not buttery inserts, set into
sockets cut into the panel back, as a supercial assessment of the exposed
panel back might suggest. Thinning had also exposed remnants of original
nails driven into the top and bottom edges, probably to secure the strips.
These would have been driven near the center line of the original edges.
One of two original rectangular beech wood (Fagus sylvatica L.)
inserts, visible on the front, had been exposed in an empty mortise at the
back (Fig. 10a, b). The insert had been used to replace a wood defect. The
adhesive did not appear to be casein, as was used to join the planks, but
animal glue.
28
Where visible within the larger gaps, the joint faces had
been inscribed with shallow Xs, either for adhesive tooth (which seems
unlikely) or perhaps to ensure adequate glue pickup from the brush.
From the evidence then, the procedure for joining was as follows:
the joint surfaces were planed; regular Xs were carefully inscribed into
each surface and mortises were chiseled; one end of each tenon was glued
with casein into one of the mortises of one plank;
29
the same glue was
applied to both joint surfaces and to the protruding tenons; the planks
were pressed rmly together and possibly rub-joined,
30
since the glue lines
are relatively thin and do not appear to have dried in a starved condition.
After drying, the desired height dimension was achieved, and the edges
were squared with narrower strips, nailed and glued to top and bottom
edges. The sides were trimmed square and straight.
Similarly, the Palmezzano would have been about twice its current
thickness. Again, as with the Mengss tenons, poplar dowels were used
to maintain rough alignment during assembly, and then the edges were
nished. Thinning had exposed some dowels. Also similarly, long spikes
remained that would have been driven straight, and with evident skill,
near the original midline of the side edges.
Insect damage
Larger panels have proportionally greater expanses of insect-prone wood.
Practical construction from whole planks would have favored greater
plank widths. For economy and practicality, critical edges of sapwood were
sometimes left in longer planks, partly because the transition line between
heartwood and sapwood is irregular for some types of wood used for pan-
els, such as the walnut and poplar used here.
Nearly every plank of the Palmezzano had variable, discontinuous
lengths of damaged sapwood.
31
In more central, critical areas, the damaged
459 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
Fi gure 10a, b
Anton Raphael Mengs, Noli Me Tangere. On
the back, an original beechwood insert (a) in
the walnut panel can be seen through an
exposed mortise. The same insert viewed
from the front (b) before treatment and in
top-raking light.
a b
wood was replaced with inserts of linden wood to within 23 mm of the
ground. No obvious adverse reactions to consolidation or wood replace-
ment appeared after three years of observation. The apparent stability
may partly be due to poplars ability to accommodate stresses because of
its resilient, brous, low-density structure.
Intermittent areas of damaged sapwood also occurred along
some joints in the Mengs. Large wood losses had occurred at the bottom
left corner (viewed from the front) (Fig. 11a, b), and plasterlike lls had
been made, extending through the panels thickness.
32
The paint overlying
the surrounding damaged wood had blistered into foldlike undulations in
some areas. This damage was probably a result of compression from the
panels weight, possibly aggravated by setting and swelling of the wet
plaster. Despite these adverse eects, the plaster was strong and well
keyed. Its contact surface was well dispersed, which probably helped to
spread stresses.
Intermittent insect damage occurred over the remainder of the
panel, but in general this damage was not a serious structural threat and
required no treatment. However, the substantial loss at the bottom corner,
covering nearly one-third of the panels width, represented weakness in a
vulnerable area. Thinning had concentrated the entire panel weight onto a
narrower cross section, one-third of which was weakened by insect dam-
age. Some provision for added strength was considered necessary to pro-
vide adequate support and prevent further loss and damage to the large
area of paint overlying the damaged wood. Walnut inserts were tted.
33
Interactions of thinned panels and nonoriginal
auxiliary supports
Obviously the Mengs was a heavy panel while still in its original state.
Display, handling, and transport called for adequate reinforcement.
Horizontal joints could be advantageous, since gravity would tend to keep
them together in compression. Standing vertically, however, destructive
460 Bre we r
Fi gure 11a, b
Anton Raphael Mengs, Noli Me Tangere. Top-
raking light shows an insect-damaged area in
the lower left corner of the front, surround-
ing a plaster ll. Note the undulations, proba-
bly from compressive strain from the panels
weight. The same area seen from the back (b)
in normal light.
a
b
bending stresses would be more likely, imposed by considerable leverages.
Any warp from a at plane would promote buckling. Such a buckling ten-
dency would pose a long-term bending stress across the joint axes, a condi-
tion greatly exacerbated by thinning.
Evidence of a previous, perhaps original reinforcement exists as
eight sets of three holes each, spaced at regular intervals across the panel
back (Fig. 9d). It was unknown what form this reinforcement would have
taken. Had documentation of the panel back prior to the recent thinning
been available, it may have provided evidence to help with subsequent
treatment decisions.
By contrast, the structure of the Palmezzano is more logical. The
chosen wood is lighter, with the planks disposed vertically. Thus, the wood
bears weight in a more natural orientation, analogous to its mechanical
role in a living tree. With the grain vertical, buckling would be a negligible
concern, even with the panel half its original thickness. Therefore, weight
does not combine with movement across the wood grain to threaten the
Palmezzanos structure as much as in the Mengs.
The Palmezzano had no evidence of original auxiliary support.
34
At least four dierent types of battens had been applied at various times
over the long period since the panel was last thinned. Finally, remedial
action took the form of short planks glued over developing splits, includ-
ing old stretcher members taken from paintings on fabric. Eventually the
panel became choked with stopgap solutions. These additions induced
severe distortions, splits, and compression damage concentrated in the
panels center. Centralized damage occurred because overall reinforce-
ments tended to concentrate bending stresses toward the middle. This fac-
tor then combined with tension and compression stress overall caused by
restraint of lateral movement.
35
The pattern of splits shows how stresses
were interrupted over the cross-grain battens (Fig. 4c).
Also, putties (or lls) had been applied to splits that had not rst
been rejoined. This and subsequent wood movement caused compression
stress and distortions in the adjacent paint. Such disguring damage is nei-
ther easily nor totally reversible. It is better not to put llers into surface
cracks if eective structural work is not done rst to underlying splits.
For both panels, attempts to atten and reinforce them have
instead tended to weaken them further. Such treatment eorts are examples
of excessive, damaging measures that have been used to meet reinforce-
ment requirements of some large panels, as well as to serve aesthetic pur-
poses. The resulting deterioration of paintings with supports shows that
those requirements must be better understood, and they must be achieved
with better methods that maintain the integrity of the panel painting.
The consequences of thinning a large panel can be critical, mainly
because a heavy weight must then be supported by a structure made rela-
tively weak while still allowing for adequate wood movement under vari-
able conditions. It is worth examining the motivations for thinning, which
have particularly serious implications for preservation of larger panels. In
general (and leaving the question of transfer procedures aside), panels may
have been thinned for several reasons, including:
1. the mistaken belief that thinning reduces the tendency to
move and warp in response to changes in MC (the reverse is
true);
36
461 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
2. to atten and smooth the back surface, a procedure usually fol-
lowed by the attachment of battens or laminates to restrain
the panel in a attened state;
3. to lighten the panel for easier handling or transport (which
may also have the opposite eect because of increased
fragility);
4. simply to take action.
It could be tempting to lighten large, heavy panels such as the
Mengs. From a strictly practical viewpoint, sheer size would be reason
enough for cutting away some wood to provide a at surface to more
easily t a new auxiliary support. To respect and conserve the entire original
object, however, it is possible, for example, to build up an even surface of
balsa wood for battens to bear upon (Buck 1962) without removing origi-
nal panel wood to achieve the same purpose. The three remaining reasons
cited above for thinning are unjustiable with respect to preservation.
The most recent thinning of the Mengs panel appears to have been
directed at obtaining a at surface to allow adhesion of relatively large
balsa planks. The use of power tools is evident from the parallel kerf marks
of a circular saw, power-planer blade marks, gouge marks chipped deep
into the irregular walnut grain, localized rasp marks, and other damages.
Despite the large scale of work that seems to justify the use of
power tools on larger panels, the use of manually controlled hand tools is
preferable. Some power tools are double-edged swords that can speed
work but also easily outstrip the intention and control of the user.
37
A
higher speed of treatment, for whatever reason, should not endanger the
painting. In this regard, responsibility for the rate of treatment and its
eects extends beyond the conservator to all custodians of cultural prop-
ertyadministrators, curators, dealers, and owners.
In an eort to prevent buckling of the Mengs, strips of slotted
metal had been screwed into the edges of the panel and balsa laminate
(Fig. 12). Obviously, even though cross-laminated, balsa did not prove
suciently rigid to prevent buckling when the panel was upright. The
metal edging provided a relatively rigid outer framework that met the
immediate reinforcement need but that had serious consequences for
the painting.
38
Unrestrained, the panel would expand and contract as a unit, the
top moving upward and downward with changes in MC. With such a large
panel, lateral movement across the wood grain could be on the order of
50 mm, if fully equilibrated over a 30% change in RH.
39
However, the
entire panel could not move as a unit. Instead, the planks were individually
constrained to expand and contract around the wood screws at each end.
At lower humidities, the panel would contract across the grain, and either
the wood had to split or the joint adhesive had to give way, depending on
whichever was weaker. Though casein is normally a strong adhesive, the
walnut wood was stronger, even across the grain, so the joints failed in
tension across the adhesive layer. They probably opened catastrophically,
as zippers sometimes do, especially if the panel was subjected to relatively
rapid and large changes in RH.
40
One joint near the center of the panel
had completely parted.
Environmental history aects the stress distribution in wooden
panels.
41
Seasoned planks develop a particular stress distribution before
being assembled. Once the planks are joined, grounded, and painted on
462 Bre we r
Fi gure 12
Anton Raphael Mengs, Noli Me Tangere. Screw
holes in the panel edge where metal reinforce-
ment strips were attached. The panel is on the
right. Note the layers of balsa/wax-resin and
fabric laminate and the saw marks in the
panel where the balsa was carelessly trimmed.
one side, a dierent overall stress distribution develops that depends on envi-
ronmental interactions. With larger panels, the total (elastic) stress in the
panel structure is accordingly greater.
42
For the Mengs, the combination of thinning and disjoining
appeared to have reduced the physical equilibrium of the individual
planks. Once disjoined, they responded to the internal stress with defor-
mation. The plank edges at the joints, originally parallel, became con-
toured to the irregular grain direction of their respective planks. Thus,
the joint gap varied by millimeters along some disjoined sections (Fig. 9d),
and the joint faces no longer met continuously or squarely. Such potential
damage from woods reaction to stress release should discourage the thin-
ning of panel paintings.
During the rejoining process, wood inserts were tted to the gaps
so that no original wood was removed.
Treatment considerations
Planning is important for any large panel treatment. The order and choice
of treatment steps should be logical in relation to the treatment as a whole
and should not foreclose later treatment options. The greater scale of treat-
ment for large panels usually makes backtracking dicult and costly. A cau-
tious, considered approach, in which each stage is tested, should be adopted.
As an example, radiographic examination of the Palmezzano did
not reveal enough of the condition of the wood, prior to treatment, to
ascertain the full extent of damage, since the battens were obscuring the
panel wood. The possibility that the panel might require extensive wood
replacement was anticipated with a more thorough facing than the panels
apparent condition warranted. Halfway through a batten or cradle removal,
with splits all around, one cannot easily move a large panel to apply a fac-
ing that should have been anticipated earlier.
Photodocumentation is important for the back of the panel, as
well as the front.
43
It is therefore necessary to have larger panels disposed
so that necessary photography can be done at any treatment stage. Even
with the best photographic resources, adequate space is required for the
necessary distances and angles. Also, large panels invite strong lighting,
especially for overall photographs, so heat eects should be considered
(Wolters and Khn 1962). Short-duration electronic ash units have a less
drying eect than the heat associated with continually lit tungsten lamps.
Though not easily achieved, relatively constant humidity should
be maintained to minimize stresses from warping movements during treat-
ment. Rejoining can take days for larger panels because thicker joints and
less-absorbent, higher-density woods require longer drying periods.
It is sometimes better to allow sucient time for the panel struc-
ture to equilibrate during treatment, to avoid stress that might precipitate
damage. When the Mengs was rejoined, for example, the balsa reinforce-
ment was removed in stages between which the exposed panel was allowed
to equilibrate to a more stable curvature (prior to rejoining) unimpeded
by restraint caused by the reinforcement or its moisture-barrier eect. If
rejoined before equilibration, joints could fail again prematurely.
It is possible to manipulate humidity to facilitate some proce-
dures. Larger panels especially can bind or lock sliding reinforcements
because of greater total movement and rigidity. It may be possible tem-
porarily to raise or lower the humidity slightly in order to loosen sliding
463 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
members of cradles for easier removal, for example.
44
This procedure
could prevent greater stress to the painting from unnecessary tool work.
Structural treatments of large panels make great demands on a
conservator and can take a long time. Assuming that they have equal abili-
ties, one conservator usually takes at least twice as long as two, and the
demands on stamina are doubled. A team benets from improved safety
and morale, and its members can help one another in making decisions,
thereby achieving quicker and better results.
Access and control
Easy access is an advantage for structural work. Access is more dicult
with larger panels since the conservator must move around the panel. If
the panel is horizontal, the conservator must nd some means of reaching
the work area, which is often in the middle of the panel. It is important to
establish a comfortable position, since the work may be of long duration
or require sustained precise and safe manipulation of tools (Fig. 13).
Horizontal support of such a large, thin, heavy panel as the Mengs
presents some problems with regard to treatment procedures and stress
distributions in the panel during extensive, prolonged structural work.
Concern arose that warp movement would be restrained by the panels
own weight if it were laid horizontally, causing detrimental bending stress.
It is dicult to judge the eect of such warp restraint, especially
in larger panels. For example, it was anticipated that once the balsa and
wax-resin were removed from the Mengs, a dierent curvature would
ensue. Laid horizontally, the panel would almost certainly have warped
away from a table surface. The suspended weight would have caused bend-
ing around the supporting fulcrum(s), with a risk of breakage at the weak-
ened joints or in worm-damaged areas. Therefore, it was considered
undesirable to treat such a panel horizontally before adequate structural
consolidation was achieved.
Alignment and rejoining are generally more dicult for larger
panels than for smaller ones. Suitable temporary supports and apparatus
must be available for operations such as rejoining. The approach must
464 Bre we r
Fi gure 13
The conservator, left, kneeling on a bridge
used to gain access to the back of a large
panel laid horizontally for structural treat-
ment of the wooden support.
meet the relative complications of treatment and may also take advantage
of the panel structure itself.
Based on these considerations, the Mengs was placed on a side
edge with silicone paper and a length of pine batten beneath. The main
reason for standing the panel vertically was to make access to both front
and back possible during structural work. This approach (which is not a
new concept) is practical in some cases.
45
To minimize restraint and allow
the panel to adjust position, it was occasionally lifted slightly at one end.
The relatively straightforward rejoining problem of the Mengs
was especially suited to a vertical orientation. The Palmezzano, in con-
trast, had a high number of fragmented splits and generally more compli-
cated treatment demands, which made a horizontal orientation preferable.
A padded support table and rejoining apparatus were designed and built to
allow all-around access and control.
Ideally, to improve access and stabilize the panel before rejoining,
all moisture barriers and restraints, including unnecessary and nonoriginal
glue layers,
46
may be removed to allow the entire panel structure to stabi-
lize. This measure may not always be possible with larger panels, where
equilibration may have to be limited to the general area surrounding the
wood to be treated. It would have been dicult to treat either panel safely
with all previous restraints removed, though such a proposal would be
more feasible for smaller panels (Brewer 1994b).
In some cases, advantage may be taken of the immobilizing eect
of previous reinforcements during their removal to maintain some struc-
tural stability while rejoining large panels. Working from the proper top of
the Mengs, the balsa laminate and most of the wax-resin were removed
from each successive plank pair to be rejoined, leaving the remainder still
covered and the next disjoin still bridged for stability. Each freshly exposed
plank pair was then left undisturbed for at least one week to allow some
equilibration of the panels curvature before rejoining. This approach
maintained greater stability while allowing adequate joining pressures to
be applied without disrupting the remaining disjoins.
Battens were removed similarly from the Palmezzano (Fig. 14).
This was done by working across the panel grain and reducing the battens
465 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
Fi gure 14
Marco Palmezzano, The Mystic Marriage of
Saint Catherine. The panel laid facedown on
a padded table during the initial stages of
removal of the previous reinforcements. The
tools are on the chair. Note that the removal
was begun from a side-grain (nearest) edge
and progressed across the panel grain.
in a step-by-step manner while leaving them intact over and beyond the next
split. For larger panels, which may tend to move substantially on release, a
cross-grain direction of removal may be adopted. The panel can move and
warp more freely across the wood grain, with less chance of twist than if
removal proceeded irregularly or from an end-grain edge. A hand-pressured
chisel (no mallet) and a dovetail saw were the only tools used.
Curvature eects and rejoining of larger panels
The convex warp (viewed from the painted side) in many panel paintings
is largely due to the development of compression set (Buck 1972:2), as
shown by both panel examples.
47
Both were rejoined with respect to the
set warp assumed by the planks after removal of the reinforcements.
Several methods might be considered to reduce such warp and maintain
the panel in a safer, more planar conguration. Though usually done for
aesthetic reasons, this would also lessen the panels tendency to buckle
under its own weight if the planks are disposed horizontally.
In the authors opinion, set in the wood of panel paintings is not
practically reversible as yet, especially in larger panels, because most meth-
ods involve extensive intervention to the wood or have uncertain long-term
eects.
48
Raising the ambient humidity provides only a temporary reduction
in warp, because the eect of set warp in a panel equilibrated to one
humidity returns if the panel is placed in a higher humidity and allowed to
fully equilibrate. The impression of an apparent reduction in warp from
raising ambient humidity is especially evident with larger panels because of
their greater total movement. However, most observers do not have the
opportunity to monitor long-term changes in larger panels under controlled
environmental conditions until an equilibrium curvature is established.
Other methods of warp reduction are possible. As the Mengs
required extensive rejoining, V-shaped inserts could be used in the joints
to counter the curvature of the planks and achieve a relatively at panel
(Fig. 15). This method has been used for panels of all sizes, sometimes for
aesthetic reasons. However, a gentle overall curvature may be less disrup-
tive to the appearance than the resulting washboard. Photographs before
and after structural treatment of the Palmezzano, in raking light, may be
compared in this respect (Fig. 16a, b).
Insertion of wedges for the purpose of attening the panel may
be considered as a last resort, whether for structural or aesthetic purposes.
Acceptable atness is partly an aesthetic concern, of course. In general,
however, it may be preferable to respect the current overall curvature, as
determined by the original panel structure and aging eects, to preserve
intact joints and the painting in unaltered form. Panels that are partially
disjoined or split, such as the Mengs, would require either breaking the
remainder of the fractured area, with serious risks to the overlying and
adjacent paint, or inserting wedges into the parted areas as desired.
Additional stress would be imposed on the remaining intact joint or sound
wood as attening pressures were applied. From an ethical standpoint,
such an option for attening is more practical and possibly more accept-
able for a complete disjoin.
For both panels, the above options for attening would all involve
protracted and serious risks. Finally, breaking of a partial disjoin having
original, intact paint above is an important ethical issue. In consideration
of these points, the set warp and its ramications were accepted in the
466 Bre we r
Mengs and the Palmezzano, and they were conserved within the limita-
tions of their current structure and condition.
Though disjoins in smaller panels may be glued and rejoined in
one operation using appropriate apparatus, larger breaks take more time,
so that glues tend to set, or go o, earlier than desired. The procedure
must be well prepared if time is a factor (see Brewer, Some Rejoining
Methods, herein).
Alternatively, using inserts, the conservator may rejoin a long
joint progressively by working along it in discrete stages.
49
If this method
is used, it is important to ensure correct alignment in all three dimensions
from the beginning. If the relative positions of the joint faces have been
incorrectly aligned and xed in the early stages of rejoining, they may
467 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
Fi gure 15
South Netherlands, Triptych of the Holy Family
and the Trinity, ca. 1510. Oil on panel, 1500
(center) 1500 15 mm thick (visual esti-
mates). Wallraf-Richartz-Museum (INV
WRM416), Cologne. An altarpiece consisting
of oak planks rejoined with V-shaped inserts
of oak, probably to reestablish an overall at
plane after the planks developed set convex
warps (viewed from the front).
Fi gure 16a, b
Marco Palmezzano, The Mystic Marriage of
Saint Catherine. The painting, in raking light
on the left, before structural work (a) and after
structural work and before retouching (b).
a b
converge or diverge as the other end of the joint is approached. Slight cor-
rections may be made by bending softer woods back into alignment at
later stages, but some distortion and stress are then built in as well.
Auxiliary support of larger panels
Despite such rare examples as the Turners mentioned above, larger panels
generally have substantial reinforcements (Fig. 6b, c). A higher ratio of
cross-grain dimension to thickness becomes a greater concern with large
panels of higher wood density. The structural implications can be greater
for certain panel structures, such as those with horizontal planks. Even
with sound joints, the Mengs tended to buckle under its own weight if
stood on its bottom edge. Most of the wood was suciently sound and
strong to withstand even quite severe bending stresses, but the joints will
always be weaker.
50
When there is already a set warp, buckling tendency
worsens as the panel warps further out of plane because of humidity
uctuations. Such uctuations would aggravate buckling even if the panel
were relatively at at a particular humidity.
So, with reference to these concerns, the nal and most challenging
diculty, as with many large panels, was supporting the Mengs in an
upright position without restricting movement too greatly. The disjoins,
the inadequate balsa laminate, and especially the metal edge strips were no
surprise considering the panels structure, weight, and thinned state. The
critical point, however, is that a relatively rigid form of reinforcement is nec-
essary nonetheless for such large, thin, heavy panels, and the panel structure
must be suciently sound to take potential stress without rupture.
The Palmezzano is also a good example of a large, thin, weakened
panel requiring overall reinforcement of a specialized type. Internal frac-
tures remain in many panels after structural treatment, partly because they
are dicult to detect, even with radiography, especially in brous, lower-
density woods such as poplar.
The inherent weaknesses of panels such as the Palmezzano can-
not be overemphasized. A sympathetic but eective auxiliary support is
necessary in such cases. Truly satisfactory reinforcement designs with
proven eectiveness are still being sought for panel paintings of this
nature, as evidenced by the increasing amount of literature on new and
modied reinforcement designs.
51
When this article was written, an auxiliary support was being
designed and tested for the Mengs. It is therefore not presented here.
However, an auxiliary support applied to the Palmezzano is described.
The support was designed to allow greater movement, reduce the risk of
further splits and damages, and give adequate reinforcement.
The design is based partly on those developed at the London
studio of the HKI (Fig. 17ac) (see Bobak, A Flexible Unattached Auxiliary
Support, herein; Marchant, Development of a Flexible Attached
Auxiliary Support, herein; Brewer 1994c). So far it is the largest ver-
sion that attempts to realize the main principle of tailored exibility.
Horizontal tapered battens and a peripheral frame were constructed
from Sitka spruce (Picea sitchensis [Bong.] Carr.), and oak uprights were
attached to the horizontal battens to form a supporting lattice. The hori-
zontals were dovetailed into the peripheral frame for strength during han-
dling. The peripheral frame extended beyond the edges of the painting,
and a surrounding border of thick card projected up to 5 mm in front of
468 Bre we r
the paint surface (Fig. 17d), to protect the painting edges from careless
handling and frame rubs (see Fig. 4a for damage from frame rubs).
The lattice was assembled with aluminum-reinforced joints and
various fasteners of brass and stainless steel. It was made as lightweight as
possible and was thinly constructed to facilitate framing. Because of its
prototypical nature, it had to be capable of disassembly to any stage, a
characteristic it retains. The battens were made of equal thickness and
then tapered to adjust their exibility to the panels strength and potential
movement. The bottom ledge of the peripheral frame was kerf-sawed for
exibility. Both battens and ledge were steam-bent to approximate the
panels overall deection when equilibrated to about 60% RH.
52
To attach the lattice, four vertical retaining strips were cut and
positioned at regular intervals across the panel back. The strips were slid
through retainers of poplar that were glued to the panel back. Potential
stresses on the retainers were spread locally with baseplates of poplar.
469 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
Fi gure 17ad
Marco Palmezzano, The Mystic Marriage of
Saint Catherine. After treatment, showing the
back (a) with reinforcement attached; the
retaining strip (b), with its stepped prole,
being slid upward one step for removal; the
same area with the strip removed (c), showing
the retainers and baseplates and the tapered
battens next to the panel surface; the lower
left corner of the front (d), showing edge pro-
tection and the bottom ledge, which is kerf-
sawed to increase its exibility.
a b
c d
The choice of positions for the retainers was based partly on a regular
distribution across the panel and partly on the location of relatively at
gluing surfaces.
Normally, the removal of sliding battens from a panel requires a
space of twice the batten length. The required space was twice the panels
height in this case and made a modied means of removing the retaining
strips desirable. It was possible to narrow the strips at intervals equal to
the vertical distance between the retainers. This allowed the strips to be
placed directly against the lattice battens (Fig. 17b, c) and between the
retainers; they were then slid down and engaged in a functional position.
53
Built thus, the structure provided adequate reinforcement while it
was still exible enough to bend with warp movement. This reduced the
risk of restraint of lateral movement by friction and locking. It protected the
edges against mishandling and accidents. While attached, the structure still
permitted examination of much of the panel back. Most of the structure
could be quickly removed to access all of the back surface except beneath
the retainers. The retainers could be removed mechanically with relative
ease because low-density wood was used. The glue used to attach the retain-
ers could be easily swelled with water and removed with spatula and swab.
54
Since the structural work was completed, the panel has been
monitored for at least two years to determine the eectiveness of the rein-
forcement and other aspects of the treatment. Due to RH variations,
changes in deection at the middle (in relation to the side edges) have
been measured at up to 30 mmabout half the deection that was
observed under a similar RH range when the panel was structurally con-
solidated but not reinforced. The two central retaining strips have shown
increased friction as the panel has become more convex (viewed from the
front), but lateral movement has not been excessively constrained, as
occurs frequently with more rigidly battened or cradled panels of this
nature.
55
The panel appears to be adequately reinforced and moves with-
out any obvious detrimental eect.
Framing, hanging, and transit
Old wooden panels are continually subject to movementprobably nearly
as much as when they were rst painted (Buck 1952; Laurie 1967:55; Klein
and Brker 1990; Mecklenburg and Tumosa 1991). Therefore, allowance
should be made in the frame for potential panel movement. Of course,
excessive frame restraint would negate any capacity of the panels rein-
forcement structure to allow for movement. Considerations related to the
frame retention of panels are similar to those relating to auxiliary support.
Many paintings do not remain in a relatively constant, well-controlled
environment. Passive controls are not always sucient, and active controls
can malfunction in even the best-maintained buildings.
Therefore, an allowance must be made by sizing the frame rabbet
for cross-grain expansion of the panel wood. Otherwise, a bound-in-
frame condition occurs as the panel expands to press on the rabbets
outer walls. Also, it is important that framing not restrict warping move-
ment with overly rigid retention. These stresses can easily break the panel
or the frame (Museum 1955:15960). Of course, competent framers allow
space in the rabbet to avoid this possibility, but the degree of panel move-
ment can be underestimated, especially in larger panels.
470 Bre we r
Whether or not they are framed, large heavy panels are probably
better supported on a plinth or base rather than hung. In either case, but
certainly if they are hung, a strong, rigid frame is an advantage for the
protection of a larger paneland not simply during handling. The panel
painting by Mengs arrived in such a frame. In contrast, the framing of the
Palmezzano was inadequate and detrimental.
When it arrived, the Palmezzano had a shallow, imsy frame that
was hung from the panelinstead of the sensible reverse arrangement
that has the panel hung by its frame. The weight of both was concentrated
on the panel by screw eyes set into one of the half-round battens of oak
that made up the horizontal members of the glued lattice.
Large panels, especially, should not be hung from such reinforce-
ments, because the weight is thereby converted to internal stresses on the
panel wood. The weight of the panel, battens, and frame had put such a
torque on the surrounding panel wood that a cross-grain tear was induced.
This probably occurred slowly, over a period of years, since the thick over-
lying ground and paint layers, though broken, show considerable plastic
deformation. Larger panels should be framed sturdily and be hung from the
framecertainly not the reverse. In consultations with the owner and pro-
fessional framers, it was determined that a more suitable frame was urgently
needed because of inherent weakness and the dangers of mishandling.
Though suciently strong, the rabbet of the Mengs frame was
not deep enough to allow for any warping movement of the panel, so that
the panel was, in fact, retained too rigidly. Before treatment this factor was
rather immaterial because the metal edge strips allowed little movement in
any direction. After conservation, though, the rabbet could be deepened,
padded, and possibly proled where it contacted the front to allow for
inherent warp and potential movement. Rabbets shaped to the contour of
the painted surface, or camber, at the panel edges help to spread the surface
of contact between panel and frame, reducing localized stresses and fric-
tion. Abraded varnish and paint are more likely on larger framed panels
because of greater movement and resulting friction. Proling may also
help aesthetically to decrease large visible gaps from the larger panels
greater warp movement.
During transit, larger panels should be supported to minimize
the eects of weight on bending. Low-density foam may be secured
around the panel to minimize bending from weight or shock loads while
allowing some wood movement. Since a packing system can seldom con-
form to large changes in panel shape, the environmentRH, shock, and
vibration, in particularshould be controlled, especially for large, thin
panels (Mecklenburg and Tumosa 1991:190; Michalski 1991:241). For the
transport of larger panels, reputable art professionals well versed in the
proper precautions may be preferred. They should be accompanied by a
qualied conservator, if possible.
Most described treatments were done while the author was an intern spe-
cializing in panel painting conservation at the the Hamilton Kerr Institute
(HKI). Thanks go to the Getty Grant Program and to the Samuel H. Kress
Foundation, New York, for funding the internship. Other treatments were
completed while the author continued at the HKI, employed as a conser-
vator and research associate, thanks to funding by the Leverhulme Trust,
Acknowledgments
471 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
London, and the Samuel H. Kress Foundation and the HKI. The author
thanks Ian McClure for his support and, above all, for allowing him free-
dom in pursuing these treatments.
1 Transverse grain direction.
2 Changes in MC and moisture gradients in wood are the primary causes of wood movement.
Skaar (1988:chap. 4) reviews the topic thoroughly. See also Panshin and de Zeeuw (1970:206).
3 Of course, wood movement as a proportion of cross-grain dimension (percent of movement
across the grain) remains the same, no matter what the panel size.
4 This statement refers mainly to the changes in dimensions and shape that accompany an RH
change prior to equilibration. Dimensions and shape at equilibrium also depend on such
things as the proportion of tangential to radial wood, the set of the wood cells prevailing from
past conditions, and the presence of preparation and paint layers that may inuence mechani-
cal restraint and the rate of moisture permeability.
5 The eect will be less if such a plank is positioned closer to the panels longitudinal-
grain edges.
6 For example, as RH rises, the uncoated panel back usually swells rst in response to a rising
MC. The expansion is resisted by the remaining panel thickness, which has not begun to swell.
If that remaining thickness is less rigid, such as in thin panels or in woods of lower density, the
force of swelling at the back will cause a deection, producing a concave warp when viewed
from the front. For the same wood density, thicker panels will be more rigid and therefore
have greater resistance to the eect of the swelling.
7 Longitudinal permeability may be 1,00010,000 times greater than transverse permeability
(Panshin and de Zeeuw 1970:217).
8 Determined by microscopic examination of a cross section.
9 Raaello Sanzio, Transguration (151720). Oil on cherry-wood panel, 4100 2790 45 mm
thick (average). Vatican Museums.
10 Marette (1961:6567) gives a frequency distribution by wood type.
11 Not shown, J. M. W. Turner, The Opening of the Wallhalla, 1842, exhibited 1843. Oil, wax, and
resin on mahogany panel, 1130 2010 10 (bevel) to 20 mm (middle) thick. Tate Gallery,
London (inv. N00533).
12 See an early use of American mahogany (Swietenia spp.) in two paintings attributed to
Rembrandts studio of the 1640s (Bruyn et al. 1989:66878). Though not particularly large
paintings, they are both on single planks and are therefore large examples in that sense.
Moreover, the planks are from the same tree, and show rather wild (very irregular) gure,
making them even more unusual.
13 The number of joints is smaller and the clamping spans are shorter and therefore less awkward.
14 Tommaso Manzuoli, The Visitation, ca. 1560. Oil on poplar panel, 4090 2485 45 mm
(original thickness). Trinity Hall, Cambridge, England. HKI treatment no. 194.
15 Vittore Carpaccio, Saint Thomas Aquinas Enthroned between SS. Mark and Louis of Toulouse,
Adored by a Youthful Donor; (above) Virgin and Child with Angels, 1507. Oil on poplar (?) panel,
2640 1710 3040 mm thick (visual estimate by author). Staatsgalerie, Stuttgart (inv. 136).
16 Edges roughly parallel to the axial, or longitudinal, grain direction.
17 The panel itself is usually called the primary support, or simply the support. A secondary,
or auxiliary, support may be dened as an original or later structure applied to the panel,
whether attached or not, to provide overall reinforcement.
18 Also known as slotted angle, such girders are found in various forms in laboratories in many
countries. They can usually be acquired in various ange widths.
19 HKI treatment no. 73. The painting is on a thin glue-based ground. The glue appears to be
casein, judging from the color, hardness, relative insolubility, and swelling characteristics of a
Notes
472 Bre we r
ground drip at the edge. It is interesting to note, in relation to the origins of this panel, that
Mengs was in a transition period at the time this painting was commissioned, having just
arrived in Rome from Madrid via Florence (Roettgen 1993:3032).
20 Such considerations did not stop others from using heavy woods for panels that were commis-
sioned from afar. Though Rubens may be cited as an example, oak was the standard panel
wood in northern Europe, so lighter woods would not have been commonly used there.
Lighter woods, mainly softwoods and poplar, were more common in Spain and Italy, and
therefore it is curious that walnut was used here.
21 Based on a density of 640 kg m
3
from Lincoln (1986:27).
22 Marco Palmezzano, The Mystic Marriage of Saint Catherine, 1537. Oil and egg tempera on
poplar panel (visual identication), 2560 1805 20 mm thick.
23 The earliest reinforcement lattice was glued to the thinned panel with casein, an adhesive
common in Italian panels of that period (Marijnissen 1985:65) and less likely to be found as a
panel adhesive in northern Europe in the same period. This observation was subsequently
strengthened by research, kindly shared with the author by P. Balch. Known as the Calzolari
Altarpiece, the painting was commissioned for the Church of S. Agostino in Cesena, near
Palmezzanos native Forli. The painting had been moved to the Ercolani collection in Bologna
by 1776. Cavalcaselle saw it in England in 1860, stating that it appeared damaged [and] comes
from the Ercolani in Bologna (Questopera non molto bella e danneggiata pervenne alla Raccolta
Ercolani di Bologna) (Grigioni 1956:575). Therefore, it seems likely that the earliest lattice and
some related damages are at least 120 years old, or probably nearly twice that age.
The lattice was constructed and then glued as a unit to the thinned, attened panel. This
was evident from the dowelled cross-halving joints of the lattice, exposed during removal. The
dowels were set into tapered holes and nished ush on the unexposed side of the lattice.
24 This technique included a thick gesso ground and a combination of oil and tempera paint.
25 Most planks were cut to the taper of the tree trunk for minimal waste, and the topmost end of
one plank was positioned beside the bottommost end of its neighbor.
26 Interestingly, the cutting direction caused by the bevel of the panel makers chisel resulted in a
distinctly buttery-shaped prole in many of the mortises, when viewed from the back. Thus,
it is now possible to mistake the mortises for original insert sockets of the buttery type, with
inserts set in from the back, a technique seen in some panels.
27 Loose tenons, probably of holm oak (Quercus ilex L.) were used to align the plank edges dur-
ing assembly. Regarding origins, both woods could be found in Italy and Spain at the time. In
Spanish panels, walnut is found mainly in panels from the regions of Navarre and Castille
(Marette 1961:68). It is possible that the panel was constructed in Spain, the painting begun
there by Mengs and nished after his move to Rome.
28 This observation was not tested chemically.
29 The tenons were tted very loosely, with at least a 3 mm gap all around. Before the joining,
the tenon glue may or may not have been allowed to dry. It is interesting that the t is quite
free, with little contact area, suggesting that the tenons were more for alignment in assembly
than for joint strength.
30 The surfaces are rubbed together to thin the glue line until the increasing adhesive strength
makes further rubbing very dicult. For such large planks, this might have been done with
mallet blows at the plank ends while joining pressure was applied.
31 Consolidation of one damaged edge has been presented in a previous article (Brewer 1994a).
32 The ller, harder than plaster of Paris, had keyed well into the surrounding damaged wood. It
had swelled on setting, a characteristic of plaster of Paris (Gettens and Stout 1966:253).
33 After a thorough facing of the area, wood inserts of similarly grained European walnut were
applied to the Mengs. Only insect-damaged wood was removed, to within 23 mm of the
ground, as with the Palmezzano. Though the wood was sized with Paraloid B72, the use of
water-based glue caused considerable swelling of the higher-density walnut, which then
tended to delaminate from the back of the weakened casein ground. It was then necessary to
remove the remaining wood to the ground, which was strengthened with a thin size, and the
inserts were directly tted.
473 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
This approach seemed to work well, though the original wood-ground interface was lost.
Epoxy llers, which would not swell the wood so much, were considered, but penetration and
ow are hard to control. Also, most cured epoxies are mechanically intractable to the move-
ment of surrounding wood. They are also too ecient as moisture barriers, and adhesion of
other glues is limited (Skeist 1977:chap. 26). Rather than the epoxy, something like a strong,
exible, two-part polyvinyl acetate or a tough acrylic, soluble in organic solvents, might be
more suitable. The behavior of the wood of the Mengs was in critical contrast to that of the
brous, lower-density wood of the Palmezzano, which swelled less from the same glue and
did not transfer the swelling detrimentally. There was no apparent eect on the strongly
adhered gesso ground of the Palmezzano.
34 It would likely have been two or more dovetail-section tapered battens, set into matching
grooves in the panel back, typical for such panels (Marette 1961:pl. 14, no. 56).
35 Such cross-grain battens, if not tted carefully to a panels surface, are usually glued to the
high spots only. Aside from the inherent restrictions on transverse wood movement from any
glued restraint, the intermittent attachment also helps to localize and concentrate tensions
due to restraint of dierential movements of the two components. Consequently, splits are
multiple and are distributed accordingly.
36 The greater tendency of thinned panels to move and warp in response to changes in MC is
due partly to steeper moisture gradients and partly to the decrease in panel rigidity. Lucas
(1963:166) referred to this, perhaps in a slight understatement, as a loss of constructional
strength. See note 6, above, for a partial explanation.
37 The classic horrifying scenario of knots being pulled out along with the paint by power
routers is sensational but entirely possible. Vibration is another concern.
38 The more informed choice of balsa makes it unlikely that those who applied the balsa rein-
forcement also attached the edge strips.
39 This estimate is based on movement of 1.8% (average of 2.0% tangential and 1.6% radial) over
an RH range of 3090% at 25 C (Building Research Establishment 1975:6).
40 See Gordons absorbing discussion of critical Grith crack length (Gordon 1978:chap. 5,
esp. 98105).
41 Because wood is viscoelastic, seasoning establishes a general stress distribution, but it does not
make timber free from stress.
42 An example of the eect of releasing elastic stress is seen in the warp that may immediately
develop as oak planks are sawn from thicker timber that has already been dried to EMC.
(These stresses are sometimes called tensions.) Paintings on oak, if recently disjoined, will
sometimes show variable gaps that may be partially or wholly due to the same reason.
43 Adequate photodocumentation of the condition and potential identifying features of the panel
back is an advantage for treatment. Examination and photography with other light sources,
such as infrared, may also reveal important historical or conservation-related information that
future events may obscure or destroy.
44 Water should not be applied directly to the panel wood because it increases the risk of com-
pression set at the back, with a subsequent tendency to greater set warp.
45 Other larger panels have been treated in a vertical position. One example is the Piet de
Villeneuve-ls-Avignon (145456) by Enguerrand Quarton (Louvre INV RF1569), also painted on
a walnut support (Bergeon 1990:3538).
46 Animal glue, for example, can form a substantial restraint.
47 Warp from this cause is modied by movement deriving from the cuts of the planks and by
any restraint caused by applied layers and by joints.
48 Some eects of attening by water application have been noted. Flattening with moisture and
pressure over an extended time to induce wood plasticization and a tension set in opposition to
an existing compression set has been discussed lucidly by Buck (1963 and esp. 1972). Though
these elements were discussed theoretically, the practical application and consequences were
not conclusive. Regarding slippage and attening, Buck states that his conclusions about slip-
474 Bre we r
page at the molecular level in panels restrained at by balsa laminates must remain theoretical
until an occasion arises to remove the balsa backing from one of the panels and to observe
the actual behaviour. (Buck 1972:11). Observations and an attempt to accomplish this process
under similar conditions have not convinced the author that attening can be achieved by
higher MC and pressure alone. Gordon (1976:143) asserts that heat is the principal agent for
bending wood. Other key elements, however, are time and whether the desired eect can be
achieved within a practical treatment period. A combination of heat and moisture applied over
an extended period would subject most panel paintings to considerable risks.
Chemical methods, through vapor exposure or impregnation (for example, see Wolters
1963), either interfere physically with the moisture response of the wood or alter the chemical
nature of the wood. Both results alter the nature of the panel painting structure in ways that
have not been tested over long periods. Again, the risks seem prohibitive. The eects and
eectiveness of attening methods should probably be investigated with controlled studies.
49 One method of rejoining uses wood inserts glued into V-section channels that are cut into the
panel back along the split or disjoin (Uzielli and Casazza 1994:21; Bergeon 1990:22).
50 Recall that even the strong original joints of casein partedrather than the surrounding
walnut wood.
51 See, for example, the increasing frequency of articles on this subject in the journal OPD
Restauro (198693).
52 This is a recommended average RH level for wooden objects (Thomson 1978:85).
53 Another possibility to facilitate the use of such long battens or similar strips for large panels
is to construct the retainers from base blocks, glued to the panel, and a removable retaining
plate screwed or bolted to threaded metal inserts in the blocks.
54 Evostik Resin W, a virtually 100% polyvinyl acetate resin, applied as a dispersion (Howells
et al. 1993).
55 As a warp ensues, a panel that is more exible than a reinforcing batten, for example, per-
forms like a exible reinforcement, so that the proper roles are reversed. Rather than the bat-
ten bending to conform to the panels warping movements, the panels warp is bent back on
itself to conform to the reinforcement. Thicker panels, being more rigid (other things being
equal), increase the friction against rigid reinforcements. When the friction exceeds the tensile
strength of the panel, the panel will break from the stress. Bending in panels involves stresses
of a more complex nature than can be discussed further here.
Evostik Resin W, Evode Ltd., Common Road, Staord, England.
Paraloid B72, Conservation Resources (U.K.) Ltd., Pony Road, Horspath Industrial Estate,
Cowley, Oxfordshire, England.
Bergeon, S.
1990 Science et patience, ou La restauration des peintures. Paris: Editions de la Runion des
Muses Nationaux.
Bodig, J., and B. A. Jayne
1982 Mechanics of Wood and Wood Composites. New York: Van Nostrand Reinhold.
Bomford, D., J. Dunkerton, D. Gordon, A. Roy, and J. Kirby
1989 Art in the Making: Italian Painting before 1400. London: National Gallery.
Brewer, A.
1994a A consolidation/ller system for insect-damaged wood. Hamilton Kerr Institute
Bulletin 2:6872.
1994b Aluminum devices as temporary helpers for panel structural work. Hamilton Kerr
Institute Bulletin 2:7376.
References
Materials and Suppliers
475 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
1994c An auxiliary support case history. Hamilton Kerr Institute Bulletin 2:6167.
Brown, C., A. Reeve, and M. Wyld
1982 Rubens The Watering Place. National Gallery Technical Bulletin 6:2639.
Bruyn, J., B. Haak, S. H. Levie, P. J. J. van Thiel, and E. van de Wetering
1989 A Corpus of Rembrandt Paintings. Vol. 3, 163542. Dordrecht: Martinus
Nijho Publishers.
Buck, R. D.
1952 A note on the eect of age on the hygroscopic behaviour of wood. Studies in
Conservation 1:3944.
1962 Is cradling the answer? Studies in Conservation 7(3):7174.
1963 Some applications of mechanics to the treatment of panel paintings. In Recent Advances
in Conservation, ed. G. Thomson, 15662. London: Butterworths.
1972 Some applications of rheology to the treatment of panel paintings. Studies in
Conservation 17:111.
Building Research Establishment
1975 The Movement of Timbers. Technical note number 38, May 1969 (revised August 1975).
London: Her Majestys Stationery Oce.
Castelli, C., and M. Ciatti
1989 I supporti lignei dei dipinti e i sistemi di traversatura: Unanalisi storica e alcune
proposte operative. In Il restauro del legno, vol. 2, ed. G. Tampone, 14154.
Florence: Nardini.
Desch, H. E.
1956 Timber: Its Structure and Properties. London: Macmillan.
Gettens, R. J., and G. L. Stout
1966 Painting Materials: A Short Encyclopaedia. New York: Dover Publications.
Gordon, J. E.
1976 The New Science of Strong Materials, or Why You Dont Fall through the Floor. 2d ed.
London: Pelican Books.
1978 Structures. London: Penguin Books.
Grigioni, C.
1956 Marco Palmezzano, pittore forlivese: Nella vita, nelle opere, nellarte. Faenza: Fratelli
Lega Editore.
Howells, R., A. Burnstock, G. Hedley, and S. Hackney
1993 Polymer dispersions articially aged. In Measured Opinions: Collected Papers on the
Conservation of Paintings, ed. C. Villers, 2734. London: UKIC.
Jobling, J.
1990 Poplars for Wood Production and Amenity. Forestry Commission bulletin no. 92. London:
Her Majestys Stationery Oce.
Keith, L.
1994 The structural conservation of Maso da San Frianos Visitation Altarpiece. Hamilton
Kerr Institute Bulletin 2:7782.
476 Bre we r
Klein, P., and J. Bauch
1990 Analyses of wood from Italian paintings, with special reference to Raphael. In The
Princeton Raphael Symposium, ed. J. Shearman and M. B. Hall, 8592. Princeton:
Princeton University Press.
Klein, P., and F. Brker
1990 Investigations on swelling and shrinkage of panels with wood support. In ICOM
Committee for Conservation 9th Triennial Meeting, Dresden, German Democratic Republic,
2631 August 1990, Preprints, vol. 1, ed. K. Grimstead, 4143. Los Angeles: ICOM
Committee for Conservation.
Laurie, A. P.
1967 The Painters Methods and Materials. New York: Dover Publications.
Lincoln, W. A.
1986 World Woods in Color. New York: Macmillan.
Lucas, A. W.
1963 The transfer of easel paintings. In Recent Advances in Conservation, ed. G. Thomson,
16568. London: Butterworths.
Mancinelli, F.
1990 La Transgurazione e la Pala di Monteluce: Considerazioni sulla loro tecnica esecutiva
alla luce dei recenti restauri. In The Princeton Raphael Symposium, ed. J. Shearman and
M. B. Hall, 14950. Princeton: Princeton University Press.
Marette, J.
1961 Connaissance des primitifs par ltude du bois. Paris: Picard.
Marijnissen, R. H.
1985 Paintings: Genuine, Fraud, Fake: Modern Methods of Examining Paintings. Brussels: Elsevier.
Mecklenburg, M. F., and C. S. Tumosa
1991 Mechanical behavior of paintings subjected to changes in temperature and relative
humidity. In Art in Transit: Studies in the Transport of Paintings, 173216. Washington,
D.C.: National Gallery of Art.
Michalski, S.
1991 Paintingstheir response to temperature, relative humidity, shock, and vibration. In
Art in Transit: Studies in the Transport of Paintings, 22348. Washington, D.C.: National
Gallery of Art.
Museum
1955 The care of wood panels. Museum 8(3):13994.
Panshin, A. J., and C. de Zeeuw
1970 Textbook of Wood Technology. Vol. 1. New York: McGraw-Hill.
Roettgen, S.
1993 Anton Raphael Mengs, 17281779, and His British Patrons. London: A. Zwemmer and
English Heritage.
Skaar, C.
1988 Wood-Water Relations. Berlin: Springer-Verlag.
Skeist, I., ed.
1977 Handbook of Adhesives. 2d ed. New York: Van Nostrand Reinhold.
Thomson, G.
1978 The Museum Environment. London: Butterworths.
477 Pracri cai As rtcrs or rnt Srrucrurai Cos trvari o or Larot Pati Pai ri os
Tsoumis, G. T.
1991 Science and Technology of Wood: Structure, Properties, Utilization. New York:
Van Nostrand Reinhold.
Uzielli, L., and O. Casazza
1994 Conservazione dei dipinti su tavola. Fiesole: Nardini Editore.
Wolters, C.
1963 Treatment of warped wood panels by plastic deformation; moisture barriers;
and elastic support. In Recent Advances in Conservation, ed. G. Thomson, 16364.
London: Butterworths.
Wolters, C., and H. Khn
1962 Behaviour of painted wood panels under strong illumination. Studies in
Conservation 7(1):19.
478 Bre we r
I saw a group of students huddled before a painting. Their noses were almost, but
not quite, touching the panel and the soon-to-be practicing conservators were eagerly
scanning the surface. Out of curiosity I approached the group and asked them what
the problem was. They started commenting on the craquelure, the pigments used,
retouches, etc. It was all technically quite sound. I asked them if they would mind
stepping back about four feet. Somewhat reluctantly they complied, and then I asked
them what they saw. There was silence. I repeated the question. One of the students
nally ventured, A painting. Of what? I asked. An angel on a hill. Exactly.
The panel in question was Flemish, some school piece of Thierry Bouts perhaps.
A delicate, svelte angel in a white, billowing gown holding a sword aloft stood
triumphant on top of an emerald-green hillock. A magical, jewellers landscape with
winding, dusty roads, Brussels Sprout-like trees, pilgrims and horsemen threading
their way through the sun-drenched countryside, and a many-turreted castle receded
into an azurite innity beyond the hillock. This meant nothing to them as far as
I could tell. The students had not started their examination by considering the
painting as a work of art, but as an object, a thing, with ailments. There was no
sympathetic attention and they may just as well have been looking over a used car.
If students are not taught rst to experience works of art as objects capable of
providing us with aesthetic pleasure, they will never be able to apply their technical
knowledge and craftsmanship in such a way that the integrity of the work and its
tradition are totally respected.
x. x. raiit, untr a ruii xoo wirn tt. tuiinio a noust or iirt
T
nt ruri rurt costrvari o srarr at the Metropolitan
Museum of Art, New York, has completed the conservation treat-
ment of the Gubbio studiolo, after more than a decade of work.
1
This essay provides a summary report of some technical aspects of the
conservation treatment of the intarsia support panels.
2
The studiolo is a
splendid example of a Renaissance study; it was built between about 1477
and 1483 for Federico da Montefeltros ducal palace in Gubbio, Italy.
Federico da Montefeltro (142282), duke of Urbino, was a wealthy and
important patron of the sciences and the arts in the fteenth century. He
commissioned numerous works of art for his palaces, including many
intarsia works and two studioli: one for his main ducal palace in Urbino,
which still exists in situ, and the other for his palace in Gubbio (Remington
1941; Winternitz 1942; Cheles 1991; Bagatin 1992; Raggio 1992). The
479
Antoine M. Wilmering
A Renaissance Studiolo from the
Ducal Palace in Gubbio
Technical Aspects of the Conservation Treatment
painter, engineer, and architect Francesco di Giorgio Martini (14391502)
directed the expansion of the ducal palace at Gubbio, which started in
1476 or shortly thereafter. The new Renaissance palace that emerged
housed the studiolo, which also must have been designed and executed
under Francesco di Giorgios supervision. The studiolo, which was proba-
bly used as a small room for study or education, has an irregular ground
plan of about 13.7 m
2
and consists of intarsia wall paneling that originally
extended from a tiled oor to a height of 2.8 m. The intarsia panels create
the illusion of an elegant interior with a trompe loeil bench and wall
cupboards containing, among other things, books, musical instruments,
Federico da Montefeltros coat of arms, his armor, and, in the central
panel, the Order of the Garter (Figs. 1, 2). A set of panel paintings
attributed to Justus of Ghent (active ca. 146080) or Pedro Berruguete
(ca. 14501505) depicting the liberal arts is believed to have been mounted
above the intarsia panels (Davies 1955:4553).
3
A spectacular gilded and
polychrome painted coered ceiling had been mounted at 5.3 m high,
supported by an equally rich decorated cornice. A Latin phrase reecting
Federicos humanist background appears in carved and gilded letters in the
frieze above the intarsia panels. The Latin text,
4
which very likely refers to
the paintings, reads:
480 Wi l me r i ng
Fi gure 1
Studiolo of Federico da Montefeltro, duke of
Urbino, from the ducal palace, Gubbio, as dis-
played in the Metropolitan Museum of Art,
New York, in the 1950s. The oor, modeled
after the fteenth-century original (ca. 1477
83), and the window surround are modern
reconstructions.
ASPICIS ETERNOS VENERANDE MATRIS ALVMNOS
DOCTRINA EXCELSOS INGENIOQVE VIROS
VT NVDA CERVICE CADANT [ORA PARENTIS
SVPPLIC]ITER FLEXO PROCVBVERE GENV
IVSTITIA PIETAS VINCIT REVERENDA NEC VLLVM
POENITET ALTRICI SVCCVBVISSE SVE.
(Menichetti 1987; Raggio 1992)
See how the eternal students of the venerable mother,
Men exalted in learning and in genius,
Fall forward, suppliantly with bared neck and exed knee,
Before the face of their parent.
Their reverend piety prevails over justice and none
Repents for having yielded to his foster mother.
5
Guidobaldo da Montefeltro (b. 1472), Federicos only son and sec-
ond duke of Urbino, died in 1508 without an heir. From 1508 to 1631 the
duchy belonged to the House of the della Rovere; when that line ended the
duchy fell into the hands of the Papal States. At that time, around 1631,
many of the artworksincluding paintings, the books from Federicos
famous library, and other portable objectswere removed from the ducal
481 A Rtai s s act Sruni oio rrox rnt Ducai Paiact i Gutti o
Fi gure 2
Detail of the central wall panel of the Gubbio
studiolo during conservation treatment. The
panel is constructed with various irregularly
shaped matrix sections. Each door, for
example, is one matrix section. The genius
of the intarsiatori can be seen in the sophisti-
cated play between light and shadow, as well
as in the intricate details of the writing uten-
sils pouch and the Order of the Garter.
palaces. The paintings were removed form the walls of the studiolo in 1673
and taken to Florence (Raggio 1996). It was not until the end of the nine-
teenth century, however, when a local family owned the ducal palace, that
such major architectural xtures as chimneypieces, door surrounds, and
decorative ceilings were removed. In 1874 the studiolo was bought by Prince
Filippo Massimo Lancellotti. He had the studiolo dismantled (except for the
paintings, which had already been removed) and moved to his villa in the
hills of Frascati, near Rome. The rst major restoration of the studiolo
took place before it was installed in Lancellottis villa.
6
A note discovered
in one of the studiolos doors conrmed the restoration and dated its com-
pletion to September 1877. In 1937 the German art dealer Adolph Loewi
purchased the studiolo from the Lancellotti family. Loewis workshop in
Venice executed the second restoration (Fig. 3).
7
In 1939 the Metropolitan
Museum of Art purchased the studiolo and displayed it until 1967. The cur-
rent conservation campaign started in 1987 with a rotating team of conser-
vators, conservation fellows, and students. The project was completed in
April 1996 and the room opened to the public in May 1996; the exhibition
included a didactic presentation about the history and conservation treat-
ment of the room.
The Gubbio studiolo was commissioned, designed, and skillfully executed
during the height of the Italian intarsia tradition, which started in the
middle of the fourteenth century and lasted roughly two hundred years.
From the second quarter of the fteenth century onward, the intarsiatori
applied linear perspective (the representation of three-dimensional space
on a plane surface) in their work and soon were given the honorary title
i maestri della prospettiva, the masters of perspective (Ferretti 1982). The
Florentines in particular had mastered the technique of creating a perfect
The Intarsia Panels
482 Wi l me r i ng
Fi gure 3
The Gubbio studiolo in the workshop of
Adolph Loewi in Venice in 1938 or 1939,
shortly after the restoration of the room had
been completed. The conguration of the
panels is not accurate. This staged setting of
the studiolo was intended to show as much of
the room as possible to prospective buyers.
trompe loeil image with naturally colored woods. The workshop of
Giuliano da Maiano (143290), who was a woodworker, architect, and
one of the most celebrated intarsiatori of the fteenth century, probably
produced the intarsias of the Gubbio studiolo (Raggio 1992). The three-
dimensional illusion of the panels results from the application of the rules
of linear perspective combined with a thorough understanding of the deli-
cate play between light and shadow. The extremely skillful craftsmanship of
the woodworkers is best illustrated with some intarsia details that reveal
the precision and subtleties of the inlay (Fig. 2).
A basic form of intarsia is called intarsia a toppo: repetitive, geometric
decorations created by inlaying complicated, often symmetrical patterns
into a walnut substrate or matrix. The designs were often simple. The
woodworkers laid them out with measuring tools such as rulers, squares,
and compasses. The more elaborate intarsia images required design draw-
ings and cartoons. Generally the painters, who often collaborated with
woodworkers on other projects as well, supplied the designs and cartoons
for gurative intarsias. Alessio Baldovinetti (142599), for example, sup-
plied a cartoon for the Nativity panel, which Giuliano da Maiano executed
for the new sacristy of the Duomo of Florence (Haines 1983).
The steps of creating an intarsia panel are not known to have
been recorded; however, examination of the various intarsias suggests that
some were made as follows: The intarsiatori rst cut the wood sections to
be inlaid according to a design or cartoon. They used saws, planes, adzes,
chisels, and knives to form these approximately 5 mm thick sections, or
tesserae, into the desired shapes. The next step was to outline, cut, and
excavate the matrix wood (usually walnut), so that the various tesserae
could be inlaid into the excavated areas. The intarsiatori typically used a
shoulder knife, rst, to set the outline of the areas to receive the inlay and,
second, to remove the wood with gouges down to the depth of the rst
knife cuts. They next made a new series of knife cuts along the same out-
line and removed more wood down to a depth of about 5 mm. Once the
matrix wood was ready for inlay, the intarsiatori secured the tesserae into
the matrix with hot protein glue or cold casein glue. After this initial
round of inlay, they planed the surface until it was level. By then, a basic
design could be recognized. The use of the shoulder knife caused the walls
of the excavated wood to taper slightly, creating a very tight-tting inlay
much tighter than that achieved with later marquetry techniques. The
matrix often formed part of the image and therefore, in many instances,
remained partly visible after the work was completed.
The intarsiatori further inlaid the panel to create ner detail,
adding rounds of inlay until satised with the nal image. They cut
slightly less deeply after each round of inlay, and each time they planed
the surface of the wood. No known cartoons for intarsias have survived, a
fact that suggests that the cartoons were cut and used during the intarsia-
making process.
8
The intarsia panels from the Gubbio studiolo were made using
these techniques. Locally available woods such as walnut ( Juglans spp.)
in various shades, pear (Pyrus spp.), mulberry (Morus spp.), bog oak and
brown oak (Quercus spp.), spindle tree (Euonymus spp.), cherry (Prunus
spp.), and others were part of the palette of the woodworkers. These
woods provided a variety of colors and shades, as well as the dierent
Intarsia Technique in the
Fifteenth Century
483 A Rtai s s act Sruni oio rrox rnt Ducai Paiact i Gutti o
grain and texture so essential in creating the extraordinarily intricate intar-
sia images. One colored wood stands out as unique among the more com-
mon wood colors. It is a green wood, stained by the fungus Chlorociboria
(Blanchette, Wilmering, and Baumeister 1992). The wood is stained in the
forest, when dead trees or branches become infected by this particular fun-
gus. The intarsiatori were quite familiar with this phenomenon, and the
use of green wood can be seen, for example, in some of the inlaid book
covers and in the feathers of the small parrot in the Gubbio studiolo.
In the Gubbio studiolo the intarsiatori assembled the various
matrix sections to form a full- or half-height wall panel. They then nailed
the matrix, sections from the front, to a backing of poplar support panels,
with handwrought nails (Fig. 4). The nails pierced the back of the sup-
port, and their tips were bent over and driven back into the wood. The
intarsiatori then concealed the nail heads with a piece of inlay. In many
instances they predrilled the location for the nails to prevent the wood
from splitting.
The conservation treatment of the Gubbio studiolo has proceeded along
two paths. One proved to be a fairly straight lane, while the other is best
described as a rugged trail with narrow passes, fallen trees, and rewarding
scenic views. The straight lane involved preserving the structural integrity
of the room, including such work as stabilizing the wall panels and ceiling
construction and consolidating loose inlay and aking paint. The rugged
trail was more challenging to tread; it involved the aesthetic decisions
necessary to preserve the visual integrity of the extraordinary fteenth-
century Renaissance room. These aesthetic decisions could be made only
in relation to a virtual mental reference collection of similar intarsia
works, as well as paintings, illuminated manuscripts, drawings and prints,
Approach to Conservation
484 Wi l me r i ng
Fi gure 4
Exploded drawing illustrating how the various
matrix sections form half a panel (in this case
panel 7 top). Each matrix section is nailed to
the support panel, and the nails are typically
hidden beneath the intarsia. Each support
panel originally had one or two vertical bat-
tens for strength. Only one original batten
remains, at the bottom of panel 6, the model
for shaping those in the drawing.
architecture, furniture, and other decorative arts. All of these were prod-
ucts of a unique moment in European history, rich in humanist interests,
scientic pursuit, and artistic expression. During this vibrant time of
curiosity and imagination, a historic consciousness emerged that was not
only new to a whole generation of nobility but also new to middle-class
merchants and artisans. Human and architectural proportions and nature
were studied in depth, as were such abstract subjects as volume, color,
light, and perspective.
Therefore, we cannot simply talk about the preservation of a room
with intarsia wall panels and a polychrome ceiling that happens to have been
built at the end of the fteenth century. The studiolo, constructed at the
height of the Italian Renaissance, was designed with great deliberation,
every component serving a purpose, and even the seemingly casual place-
ment of the tesserae was carefully considered. The studiolo strongly reects
the zeitgeist of the Renaissance. During the current conservation treatment,
the goal of maintaining the integrity of the intarsia wall panels and poly-
chrome ceiling has been at least as important as the physical preservation of
the material. The aesthetic pleasure that Federico and his son Guidobaldo da
Montefeltro must have felt upon entering the studiolo is what we should be
able to feel today. As Talley says, no object should be considered solely as a
thing, with ailments. His description at the beginning of this article of the
generic Flemish landscape painting as a magical, jewellers landscape cap-
tures the essence of every work of art (Talley 1992).
9
With these aesthetic considerations foremost, the conservation
treatment of the Gubbio studiolo has proceeded; requirements have ranged
from cleaning, consolidating, and retouching the intarsia and polychrome
paint to fabricating complicated replacements for both the intarsia panels
and the polychrome ceiling components. The focus of this article is the
treatment of the supports of the intarsia panels and the coered ceiling.
10
The main concept of the conservation treatment can be summa-
rized as follows: to preserve and restore the fteenth-century character of
the studiolo. All the original elements of the room were to be conserved
11
and the nineteenth- and twentieth-century restorations kept, where
possible. These later restorations were respected as part of the history
of the studiolo; even so, they were replaced in areas where the initial
fteenth-century intention of the intarsia panel had been misinterpreted,
and the restorations had consequently disgured the image. The intarsia-
tori executed original intarsia panels with a sophisticated sense of the
delicate play between light and shadow and with a superb eye for detail.
Today the aged wood still displays more contrast and a warmer tone scale
than many of the later restorations, which have discoloredcompeting
with, rather than complementing, the fteenth-century elements.
12
Much
of the treatment, therefore, consisted of integrating past restorations to
bring out a coherence that had been compromised, within the intarsia
panels and between the intarsia panels and the polychrome elements.
New additions were kept to a minimum, and where possible they were
made reversible. Unfortunately, the polychrome paint of the ceiling ele-
ments had sustained considerable damage over time, and the later restora-
tions had badly discolored and aked. These previous restorations were,
therefore, completely removed. This removal prompted extensive repaint-
ing, which was possible because of the repetitive decorative pattern of
the ornamentation.
485 A Rtai s s act Sruni oio rrox rnt Ducai Paiact i Gutti o
The intarsia panels, and indeed the entire room, had sustained damage
from the studiolos four-hundred-year tenure at the Gubbio palace, espe-
cially in those years when the palace was neglected and abandoned. The
ducal palace housed a candle factory near the end of the nineteenth cen-
tury. Paul Laspeyres, who saw the studiolo in its original location in 1873,
described it nine years later as being in a severely deteriorated state.
13
The Lancellotti and Loewi restorations had aged, and many of their inter-
ventions had become visible. Woodworm infestation had substantially
deteriorated the supports, and they had lost structural strength.
14
In areas,
the back panels and matrix sections had separated, and in a number of
locations, the inlay was loose and protruding from the matrix sections.
Also, many of the restorations were discolored. In some instances wood
replacements had been selected without respect for either grain direction
or the proper species. Thin rosewood (Dalbergia spp.)
15
veneer, for
example, was used to restore areas that should have been restored with
brown oak or bog oak.
The intarsia images were cleaned with a variety of gentle cleaning emul-
sions containing hydrocarbon solvent, water, and soap.
16
A thin layer of
7.5% shellac was applied to the surface to saturate the wood colors and
serve as a retouching varnish.
17
Intarsia elements that had become
detached were reglued with traditional warm protein glue (hide glue).
Discolored restorations were toned with either watercolor or dilute
Golden acrylic color to create a balance with the aged fteenth-century
intarsia. Missing elements were replaced with wood, which was carefully
selected with a concern for the proper species and for similarity in texture,
grain direction, hue, and density.
A few of the intarsia images had no back supports and needed
elaborate intervention to restore their structural strength. The state of
each detached intarsia varied from panel to panel. Some panels had no
remaining hardware at all, while in others the original nails had been
clipped, and stubs ranging from 0.5 to 1.0 cm in length protruded from
the back of the matrix sections. The intarsia panels that still possessed
their original supports had survived well over the last ve hundred years
because of the exibility inherent in the original nailing system. Therefore,
it was of particular importance to restore the original nailing system in
each of the damaged panels. A number of solutions were devised to
ensure that the original pull, or force of the nails, in each panel was
approximated as closely as possible. Most boards had little, if any, planar
distortion, or warping. Existing splits and gaps were not lled or otherwise
treated, since the panels were in equilibrium with the matrix sections, and
it was important to avoid introducing any new forces.
The most eective solution to restoring the original nailing sys-
tem in the damaged panels was also the simplest, as those nails where a
stub of about 1 cm had been left could be cut with a positive thread.
Solid brass extensions were then fabricated;
18
they were hollow on one
end, which was tapped with a negative thread to t the threaded nail
stub. The other end of the brass extension was cut with a thread that
could be used to fasten it with a washer and nut to the back of the new
support (Figs. 58).
Treatment of the
Intarsia Panels
Condition of the
Intarsia Panels
486 Wi l me r i ng
Some nail stubs were too small (shorter than 0.5 cm) to be
threaded and therefore needed a dierent extension system. A hollow
piece of threaded brass, similar that used by electricians, was secured to
the nail stub with carvable epoxy resin (Araldite AV 1253/HV 1253).
Before the resin was applied, the wood surrounding the nail stub was iso-
lated with a thin layer of protein glue. The nail stubs were notched and
degreased for better adhesion with the epoxy resin. After being secured to
the matrix sections, the brass extensions were fastened to the supports by
washers and nuts (Fig. 5).
A third method was necessary in areas where the nails had been
removed completely. Small round cylinders of wood, measuring about
1.6 cm in diameter and 1.4 cm high, were glued to the matrix sections next
to where the nails had been removed. This was done to approximate as
closely as possible the original forces in the intarsia panel. The grain of
487 A Rtai s s act Sruni oio rrox rnt Ducai Paiact i Gutti o
A B C D E
Fi gure 5, above
Test and demonstration model of a variety of
attachment systems considered for the intarsia
matrix sections and the new poplar support.
From left to right: (a) an imitation of a
clipped nail; (b) a short notched nail extended
with a threaded tube glued with epoxy resin
onto the stub; (c) and (d) small round pieces
of wood glued with hide glue to the matrix,
with their grain in the same direction as the
matrix sections, protected from splitting by
small collars; and (e) a nail stub cut with a
thread and tted with a brass extension.
Fi gure 6, above ri ght
Detail of Figure 5 showing the two most
frequently used attachment systems (d and e).
Fi gure 7
Detail of Figure 5. A poplar board is attached
with the systems shown in Figure 6.
Fi gure 8
Side view of the attachment of the poplar
board show in Figure 7 (attachment system e).
From left to right: the end of a threaded brass
extension, the poplar support, the walnut
matrix with a remaining nail stub, and a strip
of inlay.
the wood cylinders was placed in the same direction to match each matrix
section. The cylinders were glued to the matrix section with hot protein
glue.
19
A plastic collar was glued around the cones with Araldite to prevent
the wood from splitting, because the supports were attached to the cylin-
ders by screws (Figs. 5, 6, and 9).
Cottonwood (Populus spp.) was selected for the new support pan-
els, to match as closely as possible the original Italian black poplar and its
properties. The wood was purchased air-dried in Louisiana and stored in
the conservation studio for two years prior to its use. The new boards,
which were mostly sawed in semiquarter direction, were abutted to
approximate as closely as possible the width of the original boards. The
fronts of all new boards were meticulously shaped to match any irregulari-
ties of the matrix section backs. This ensured that the matrix sections had
level surfaces once the new supports had been installed (Figs. 10, 11).
One board in panel 910 top had to be removed from the support
because it was too deteriorated to provide adequate structural strength for
the intarsia panel (Figs. 12, 13). X radiographs conrmed extensive wood-
worm tunnels that former restorers had lled with stucco, a plasterlike
material (Fig. 14). The board was removed, as much as possible in one
piece, so it could be kept and stored separately from the studiolo. The
remaining nails
20
attaching the matrix to the support were straightened,
and the entire board was lifted from the matrix sections. Two pieces of
cottonwood, cut to the size of the old board, were glued together to make
a new board. The old nails were reusedbut not in the traditional man-
ner, which might have broken them. They were cut with a thread so that
they could be fastened with a washer and nut through the new board.
Where necessary, additional round sections of wood were glued to the
matrix sections, in close proximity to the old nails, to ensure that there
were ample areas of attachment. The new board provided enough strength
488 Wi l me r i ng
Fi gure 9
Reverse of panel 8 top. The separate matrix
sections are clearly visible. Remaining nail
stubs have been threaded, and wooden
cylinders, with their collars, have been put
into place.
Fi gure 10
Reverse of panel 8 top. The new boards have
been attached with a combination of attach-
ment systems (d and esee Fig. 5).
Fi gure 11
Proper front side of the new support of panel
8 top. This side of the support has been
shaped to accommodate irregularities in the
matrix sections, thus ensuring that the front
of the panel (matrix and inlay) produces a
level surface.
489 A Rtai s s act Sruni oio rrox rnt Ducai Paiact i Gutti o
Fi gure 14a, b
Reverse of panel 910 top. The deteriorated
board has been removed, and the remaining
nails piercing the matrix sections have been
straightened (a). The X radiograph (b) reveals
the extensive stucco fills, which, combined
with the deterioration, were the main reason
for the removal of the board.
Fi gure 12
Panel 910 top. The panel has been pho-
tographed on one of the specially designed
project worktables. The working surface of
the tables can be tilted vertical (as shown) to
allow proper viewing of the work in progress.
Fi gure 13
Reverse of panel 910 top. The second board
from the bottom was too deteriorated to pro-
vide adequate support.
a b
to the intarsia panel so that none of the adjacent boards, which had frag-
mentary deterioration at the sides, required removal (Fig. 15).
The polychrome coered ceiling, in keeping with fteenth-century practice
and similar to the ceiling in the Urbino studiolo, had been constructed from
poplar (Populus spp.) with very little wood joinery but with an abundance of
handwrought nails (Figs. 16, 17) (Luchinat 1992:2327; Rotondi 1973).
The nailing system of the ceiling contributed to the fairly well
preserved structure of the ceiling components. Areas of extensive former
woodworm infestation, however, needed conservation treatment. The ceil-
ing had been restored and expanded with r, although the original wood
was poplar. The nineteenth- and twentieth-century polychrome restora-
tions were badly discolored and aking, while the fragmentary remaining
fteenth-century paint was fairly well preserved under a layer of grime.
The infested areas of the ceiling components needed to be treated in order
to preserve the ceiling and to ensure safe display at a height of 5.3 m.
Consolidation with synthetic resin was considered but not executed because
this plan would have substantially increased the weight of the ceiling.
Instead, a mechanical system was devised to support the infested areas from
above the polychrome hexagons. Steel plates of the proper shape were
welded to a 20 cm piece of threaded steel.
21
These plates were mounted
above the hexagons, with their thread through the backing. The nineteenth-
century beams bore the weight by means of smaller aluminum crossbars.
Treatment of the Ceiling
and Polychrome Elements
Ceiling and Polychrome
Elements
490 Wi l me r i ng
Fi gure 15
Reverse of panel 910 top. The new board is
in place, and new battens have been attached.
The substantial damage to the lower board
was not treated; the gap in the center was
lled, however, with sections of balsa placed
without adhesive.
The original fteenth-century paint was consolidated with sh
glue and the surface lightly cleaned with saliva. Most of the nineteenth-
and twentieth-century restorations were removed with either a methylcel-
lulose gel or an acetone gel, according to which binding media was used in
the later restorations. A new ground of gesso was applied after the wood
had been prepared with glue size. The decorative elements were repainted
with gouache and dry pigments in Arkon P90 resin
22
as a binder. New gild-
ing was applied in the traditional manner. All new inpainting was executed
to match the aged, original fteenth-century paint.
Through the conservation treatments discussed above, this Italian
Renaissance masterpiece has regained some of its former glory (Figs. 18, 19).
491 A Rtai s s act Sruni oio rrox rnt Ducai Paiact i Gutti o
Fi gure 16
View of the small ceiling from the window
niche during the conservation treatment. This
portion of the ceiling was almost entirely
repainted in the nineteenth century. The deco-
rative borders, with their fteenth-century
gilding and azurite paint, are mostly original.
Fi gure 17
X radiograph of the ceiling of the window
niche, showing the absence of joinery and
the abundant use of nails. The fteenth-
century paint has survived only fragmentarily,
as can be seen, for example, in the octagons,
which have dark islands of slightly denser
original paint.
492 Wi l me r i ng
Fi gure 18
Main ceiling of the studiolo after conserva-
tion treatment.
Fi gure 19
The Gubbio studiolo after conservation
treatment.
The author dedicates this article to Charles D. Wright and John Kitchin,
retired chief conservation ocers of Furniture and Woodwork at the
Victoria and Albert Museum, London, and to Bertus F. Boekho, retired
senior furniture conservator at the Historical Museum of Amsterdam, for
so kindly and generously handing him the tools of his profession.
The conservation treatment of the Gubbio studiolo could not
have been achieved without the support, advice, and interest of a number
of key players. The author would like to express his gratitude to Olga
Raggio, Iris and B. Gerald Cantor Chair of the Department of European
Sculpture and Decorative Arts, for her guidance and continuous support of
the conservation treatment of the Gubbio studiolo. He is grateful to Tony
Frantz, conservator in charge of the Sherman Fairchild Center for Objects
Conservation, for his trust and encouragement during the many years of
conservation work. He also owes a great debt to George Bisacca, conserva-
tor at the Sherman Fairchild Center for Paintings Conservation, who gen-
erously shared his vast knowledge of Italian intarsia, woodwork, and
technology. Over the years a number of conservators, conservation fel-
lows, and students have been part of the Gubbio conservation team. The
author would especially like to thank and acknowledge Susan Klim, for-
merly associate conservator, and Mechthild Baumeister, associate conser-
vator, as well as John Canonico, Rudy Colban, Mark Minor, Fred Sager,
Pavol Andrasko, Albert Neher, Dennis Degnan, Ralph Stoian, Birgitte
Uhrlau, Carmen Chizzola, John Childs, Jack Flotte, Susan Mller-Arnecke,
Constanze Doerr, Ann MacKay, Anke Tippmann, Henritte Bon-Gloor,
Carole Hall, Perry Choe, Hong Bae Kim, Amy Kalina, Stephanie
Massaux, and Jacqueline Blumenthal for their valuable contributions to
the project and for their excellent and skilled conservation work on the
Gubbio studiolo. He also sincerely thanks Bruce Schwarz and Bob
Goldman of the Metropolitan Museums photo studio for their superb
photography and print work.
1 At the time this paper was presented in spring 1995, the conservation treatment was in
progress; it has since been completed. The room opened for exhibition in May 1996.
2 A Metropolitan Museum of Art Bulletin on the Gubbio studiolo authored by Olga Raggio and
Antoine M. Wilmering was published in spring 1996 to celebrate the studiolos reinstallation.
Olga Raggio is the Iris and B. Gerald Cantor Chair of the Department of European Sculpture
and Decorative Arts. A major book on the subject is being prepared by the same authors; it is
scheduled for publication by the Metropolitan Museum of Art in 1998.
3 Two paintings of the set, Music and Rhetoric, have been preserved at the National Gallery in
London. Two more paintings, Astronomy and Dialectic, were preserved up to World War II at
the Kaiser Friedrich Museum in Berlin. The liberal arts were commonly, although not exclu-
sively, grouped as seven in the trivium and quadrivium. It is unknown whether any more paint-
ings of the group exist.
4 The Latin text had suered losses over time and was restored on several occasions. In the vari-
ous descriptions by Dennistoun (1909), Laspeyres (1882), and Gabrielli (late sixteenth century),
published in Menichetti (1987), dierent losses and discrepancies are apparent.
5 The author is grateful to John Marincola, associate professor at Union College, Schenectady,
New York, for his suggestion for a missing section in the Latin inscription, as well as for his
suggestions for the translation of the text, which is partially based on the Codice Gabrielli cited
by Menichetti (1987), Nachod (1943), and Laspeyres (1882). The translation is taken from
Raggio (1996).
Notes
Acknowledgments
493 A Rtai s s act Sruni oio rrox rnt Ducai Paiact i Gutti o
6 Paul Laspeyres, a German architectural scholar who visited the ducal palace in 1873, mentions
that Prince Lancellotti purchased the studiolo for L 7,000 and that it had been thoroughly
restored (Laspeyres 1882).
7 The author is grateful to Mrs. William J. Robertson, who shared much information on the
restoration of 1937. She was eighteen years old at the time the studiolo was at her fathers
workshop, and she recalls having been involved in the restoration of the incomplete Latin text.
The workshop operated separately from Adolph Loewi, but according to Mrs. Robertson, it
executed all restorations for the rm.
8 Intarsia making typically involves a design drawing from which cartoons on paper are pro-
duced (Haines 1983). These cartoons are suitable for transferring the design onto the wood. In
this process the cartoons are cut into smaller pieces and glued to the wood surface. This tech-
nique allows the intarsiatori to cut accurately along the outline with woodworking tools to
produce properly shaped tesserae. The technique, in which the cartoons are destroyed, is prac-
ticed today by marquetry cutters (Ramond 1989).
9 The author owes a great debt to M. Kirby Talley Jr. for kindly allowing him to reproduce the
passage quoted at the beginning of this article (Talley 1992).
10 See note 4 above.
11 Some elementsfor example, one of the boards of the support panel opposite the studiolos
entrancehad to be replaced because they no longer provided adequate structural strength.
12 The natural wood colors would have been richer, and the designs of the intarsia panels would
have had more contrast in the fteenth century. Wood owes much of its color to the gums and
deposits it contains. Light-colored woods generally have fewer of these materials than darker
colored woods. During aging, two factors play a role in the change of a woods color. First, the
gums and deposits tend to fade, much as do natural textile dyes. Second, the main components
of wood, cellulose and hemicellulose, bleach upon aging, while lignin darkens. Thus, the aging
process causes the wood colors to draw together in tone and display a less vivid chroma.
13 Noch sah ich dasselbe, wenn auch im Zustande arger Verwahrlosung im Jahre 1873
(Laspeyres 1882:77).
14 No signs of active woodworm infestation marked any of the panels or ceiling components. It
is very likely that the panels and ceiling were fumigated around 193739.
15 South American rosewood would not have been available in Italy in the third quarter of the
fteenth century. Small quantities of tropical woods may have been available through the trade
routes in Africa and Asia. It is unlikely, however, that these precious woods would have been
used in secondary areas in the intarsias (Baxandall 1986; Meilink-Roelofsz 1962; Origo 1985).
16 The mildest cleaning emulsion consisted of 600 ml Shellsol 71, 100 ml water, and 0.75%
Brij 35, a nonionic soap. The author is grateful to Richard Wolbers, associate professor in the
Art Conservation Department at the University of Delaware, for his advice in making this
emulsion. Where necessary, a slightly stronger cleaning agent (composed of 445 ml benzene,
40 ml oleic acid, 15 ml triethanolamine, and 500 ml water) was used.
17 A 7.5% shellac solution was preferred to a B72 solution, because the shellac provided fuller
color saturation for proper evaluation of the intarsia images. It formed a base for inpainting
some of the nineteenth- and twentieth-century restorations. It also protectively coated the
wood surface during consolidation in case of glue spillover.
18 The brass extensions were fabricated by Gerard Den Uijl, supervising maintainer of the
machine shop at the Metropolitan Museum of Art.
19 A high-quality protein glue with a strength of about 640 g was used. It is a very pure glue,
possessing a high shear factor and no additives, made to the specications of William Monical,
violin maker and restorer. The author is grateful to Stewart Pollens, associate conservator of
the Department of Musical Instruments at the Metropolitan Museum of Art, for advice about
this glue and its properties.
20 Many nails had already been removed, probably by the Loewi restoration of 1938.
21 The stainless steel plates were made by Gerard Den Uijl, supervising maintainer of the
machine shop at the Metropolitan Museum of Art.
494 Wi l me r i ng
22 Arkon P90 is a synthetic resin that dissolves in Shellsol 71. It is a very stable resin and has little
tendency to cross-link, or discolor, when mixed with a small quantity of Tinuvin 292, a UV
inhibitor (Rie and McGlinchey 1990).
Araldite AV 1253/HV 1253, Industrial Sales Association Inc., 39 Henry J. Drive,
Tewksbury, MA 01876.
Arkon P90 resin, Conservation Support Systems, P.O. Box 91746, Santa Barbara, CA 93190.
Brij 35, Sigma, P.O. Box 14508, St. Louis, MO 63178.
Golden acrylic, Golden Artist Colors Inc., 188 Bell Road, New Berlin, NY 13411.
Shellsol 71, Shell Solvents, 200 Pickett District Road, New Milford, CT 06776.
Tinuvin 292, Conservation Support Systems.
Bagatin, P. L.
1992 Studiolo. In Piero e Urbino, Piero e le corti rinascimentali, ed. P. Dal Poggetto, 35660.
Venice: Marsilio.
Baxandall, M.
1986 Schilderkunst en leefwereld in het quattrocento. Nijmegen: SUN.
Blanchette, R. A., A. M. Wilmering, and M. Baumeister
1992 The use of green-stained wood caused by the fungus Chlorociboria in intarsia
masterpieces from the 15th century. Holzforschung 46(3):22532.
Cheles, L.
1991 Lo studiolo di Urbino: Iconograa di un microcosmo principesco. Ferrara: Franco
Cosimo Panini.
Davies, M.
1955 Early Netherlandish School. London: National Gallery.
Dennistoun, J.
1909 Memoirs of the Dukes of Urbino. Annotated by E. Hutton. 3 vols. New York: John Lane.
Ferretti, M.
1982 I maestri della prospettiva. In Storia dellarte italiana, vol. 2, ed. F. Zeri, 457585.
Torino: Einaudi.
Haines, M.
1983 The Sacrestia delle Messe of the Florentine Cathedral. Florence: Cassa di
Risparmio di Firenze.
Laspeyres, P.
1882 Die Baudenkmale Umbriens. 9. Gubbio. In Zeitschrift fr Baukunst, vol. 31, 6282.
Berlin: Ministerium der entlichen Arbeiten.
Luchinat, C. A., ed.
1992 I restauri nel Palazzo Medici Riccardi. Milan: Silvana.
Meilink-Roelofsz, M. A. P.
1962 Asian Trade and European Inuence. The Hague: Martinus Nijhof.
Menichetti, P. L.
1987 Storia di Gubbio. Vol. 1. Citt di Castello, Italy: Petruzzi.
References
Materials and Suppliers
495 A Rtai s s act Sruni oio rrox rnt Ducai Paiact i Gutti o
Nachod, H.
1943 The inscription in Federigo da Montefeltros studiolo in the Metropolitan Museum.
Medievalia et Humanistica 2:98105.
Origo, I.
1985 De Koopman van Prato. Amsterdam: Contact.
Raggio, O.
1992 Lo studiolo del Palazzo Ducale di Gubbio. In Piero e Urbino, Piero e le corti rinascimentali,
ed. P. Dal Poggetto, 36165. Venice: Marsilio.
1996 The liberal arts studiolo from the ducal palace at Gubbio. Metropolitan Museum of Art
Bulletin 53(4):535.
Ramond, P.
1989 Marquetry. Newton: Taunton Press.
Remington, P.
1941 The private study of Federigo da Montefeltro. Metropolitan Museum of Art
Bulletin 36(1):313.
Rie, E. de la, and C. W. McGlinchey
1990 New synthetic resins for picture varnishes. In Cleaning, Retouching, and Coatings,
ed. John S. Mills and Perry Smith, 16873. London: International Institute for the
Conservation of Historic and Artistic Works.
Rotondi, P.
1973 Ancora sullo studiolo di Federico da Montefeltro nel Palazzo Ducale di Urbino.
In Restauri nelle Marche: Testimonianze acquisti e recuperi, Urbino, Palazzo Ducale,
29 giugno30 settembre, 1973, 561604. Urbino: Soprintendenza alle Gallerie e Opere
dArte delle Marche.
Talley, M. K.
1992 Under a full moon with BB: Building a house of life. In Museum Management and
Curatorship 11:34773. Oxford: Butterworth-Heinemann.
Wilmering, A.
1996 The conservation treatment of the Gubbio studiolo. Metropolitan Museum of Art
Bulletin 53(4):3656.
Winternitz, E.
1942 Quattrocentro science in the Gubbio study. Metropolitan Museum of Art
Bulletin 1(2):10416.
496 Wi l me r i ng
Probably there is no construction that suers more seriously as a result of the
movement of wood than the paint on a painted panel.
r. n. tucx, s,z
I
a roori cii xari ztn xustux or during transit, it is crucial to
control continuously the moisture content of humidity-sensitive
objects such as wood, fabric, and paper.
The use of microclimate boxes to protect vulnerable panel paint-
ings is, therefore, not a new phenomenon of the past two or three decades.
Rather, it has been a concern for conservators and curators to protect
these objects of art at home and in transit since the end of the nineteenth
century. The increased number of traveling exhibitions in recent years has
heightened the need to protect paintings during circulation (Thomson
1961; Mecklenburg 1991).
Departures from the usual climatological surroundings may cause
swelling or shrinkage of a panel, resulting in cracks, splits, and cleavage of
the support or between the support and image layers (Stolow 1967). Early
research in packing has covered some aspects that are used as criteria for
the microclimate boxes (Stolow 1965, 1966, 1967).
1
Although there may
not be an ideal relative humidity (RH) for museums, it is evident that
some objects require, or would benet from, separate microenvironments,
regardless of the chosen RH set point (Erhard and Mecklenburg 1994).
The use and design of microclimate boxes have been evolving since
1892. These boxes may be divided into three broad groups: those using an
active buer material to stabilize the internal RH, a more recent box con-
taining no added buer material, and, in recent times, boxes with an altered
gas content. Another concern is the appearance (aesthetics) of the box.
The cross-grain instability of wood has been a perennial problem to arti-
sans as it is in the nature of wood and wooden objects to seek an equilib-
rium between internal moisture content and that of the surrounding
atmosphere (Fig. 1a, b) (Buck 1961).
2
Examination of the hygroscopic behavior of various wood species
shows that green as well as old wood responds to changes in humidity
(Buck 1952, 1962).
3
The swelling and shrinkage of two panels was
Wood as a Hygroscopic
Material
497
Jrgen Wadum
Microclimate Boxes for Panel Paintings
measured with strain gauges and recorded. The investigation showed that
the movements of a new oak panel and a panel from the seventeenth cen-
tury were analogous (Klein and Brker 1990).
Experiments with beech (hardwood) and Scotch pine (softwood)
demonstrated that the hardwood has a slightly higher moisture change
rate than the softwood, and that the movement of beech samples was
therefore larger than that of the Scotch pine samples (Stevens 1961).
The ratio of the area of exposed surface to the volume of the
wood also inuences the reactivity of the wood. Thin pieces of wood
respond more quickly than thick ones, while small pieces respond more
quickly than large pieces of equal thickness. When a panel is thinned, as
is often done during the cradling process, the ratio of exposed surface to
wood is sharply increased; therefore, the diusion of moisture throughout
the bulk of the panel and the response to changes in the atmospheric envi-
ronment are accordingly accelerated.
It has also been demonstrated that the higher the temperature, the
more rapid the rate of moisture transfer. A piece of wood comes to equi-
librium about twice as fast at 24 C as at 12 C because the vapor pressure
of water at 24 C is twice as great as at 12 C, if the RH is constant.
Finally, the greater the change in RH, the faster the rate of mois-
ture transfer (Buck 1961, 1979).
The preparation of a panel before the painting process must also
be considered (for a discussion of historical techniques, see Wadum,
Historical Overview of Panel-Making Techniques, herein). The size and
ground may contain hygroscopic materials, such as glue, that also react to
changes in RH and temperature.
4
The behavior of a number of materials found in traditional paint-
ings has been analyzed under the stress of temperature uctuations and
varying RH (Buck 1972; Mecklenburg and Tumosa 1991). Another impor-
tant result of climatological uctuations is the changing stiness of paint-
ing materials and mediums in traditional paintings (Michalski 1991).
Changes in RH produce measurable changes in the dimensions of
a panel. Research has also shown that paintings change dimensionally as a
consequence of temperature, independent of a change in RH (Richard
1991). However, bearing in mind that the thermal expansion of a panel
enclosed in a case is small, the conservator should concentrate on keeping
the moisture content of the wood constant and thus ensure dimensional
stability of the panel.
5
The unanimous advice given by various authors
498 Wadum
a Air
100
80
60
40
20
0
0 10 20 30 35
20
15
10
6
4
2
1
Temperature (C)
R
H
(
%
)
g
/
k
g
g
/
k
g
b Wood
100
80
60
40
20
0
0 10 20 30 35
200
180
160
140
120
100
80
60
Temperature (C)
R
H
(
%
)
Fi gure 1a, b
Correlation between RH, temperature (C),
and grams of water per kilogram (g/kg) in
(a) air, and (b) wood.
holds that a narrow range of temperature and RH change is advisable for
the preservation of a panel painting.
Thomsons studies on the dierent properties related to RH variation
with temperature in cases containing wood set the standards for the eld
(Thomson 1964).
Calculations show that equilibrium moisture content (EMC) is
more relevant than RH, since in the microclimate box, the ratio of wood
to air will exceed 1 kg of wood per 100 l of air, a ratio that is critical to
controlling the humidity of the wood.
6
Stolow, in particular, provided much useful information and experi-
mental data on tests on enclosed packing cases (Stolow 1965).
7
Stolow,
Thomson, and Padeld were primarily interested in stabilizing RH at a con-
stant temperature (Thomson 1964, 1977; Padeld 1966; Hackney 1987).
Apart from Thomsons calculations and experiments showing the RH and
temperature changes within cases, as well as the relationships between
them, Padelds contribution to the understanding of the phenomena inside
small closed areas must be regarded as part of the standard literature.
If much wood is present, its moisture content determines the RH
of the entire volume of the microclimate box. It has been emphasized that
the diusion of water vapor through the case materials and through stag-
nant air in gaps should be kept in mind when a hermetically sealed case is
created (Padeld 1966; Brimblecombe and Ramer 1983). Padeld remarks
that water vapor diuses through air almost twice as fast as oxygen and
nitrogen and very much faster than dust particles.
8
Objections have been raised about the exhibition of objects in
almost-closed containers, because of the danger of condensation forming
on the glass or object when the temperature suddenly falls. However,
Padelds calculations and experiments conrmed that the stabilizing
eect of absorbent materials, such as the wooden panel itself, prevents
condensation. Padeld concludes that the conservation of wooden objects
in rooms that are heated but not air-conditioned often demands an
articially raised RH in individual showcases. To this end, he recommends
using saturated salt or a solution of sodium bromide to stabilize the RH
of a showcase.
Toishi describes the common belief that a closed package contain-
ing a large quantity of wood dries out when the temperature is raised,
even though the wood gives out moisture to balance the dryness of the
air. He counters, however, that the quantity of moisture vapor released
from the wood when temperature rises is generally so great that it
increases the RH (Toishi 1961).
Stolow describes the relationship between EMC and RH, as well
as the variations in RH and temperature in sealed cases containing wood.
A case at 20 C with an initial RH of 50% will increase to 53.5% RH when
the temperature is increased to 30 C. If, on the contrary, the temperature
were lowered to 10 C, the nal RH would be 46.5%. If the case were not
sealed or the air volume were very large, however, he recommends that
the internal RH be stabilized with silica gel (Stolow 1967).
To this end, Weintraub tested ve dierent types of silica gel
(Weintraub 1981; Stolow 1967). The tests showed no direct relationship
between the actual moisture content of a particular sorbent and its relative
ability to control the RH of a showcase.
9
Microclimate
499 Mi crocii xart Boxts ror Pati Pai ri os
Miura examined sorbents for their static and dynamic charac-
teristics, to estimate their ability to buer RH changes in a showcase
(Miura 1981).
Wood heated to 30 C lost 2% of its moisture content, which the
silica gel or Art-Sorb could easily absorb in order to maintain the RH at
stable values (Hackney 1987; Kamba 1993; Wadum et al. 1994).
Sealing a show-case to prevent diusion and convection and to resist, or
deform under, pressure changes up to 0.5 mb would very much reduce the
leakage of air and be a major contribution to the conservation of a wide
variety of art objects, Padeld wrote in 1966. This concept, as shall be
seen, has been a concern since the end of the nineteenth century.
In deciding the ways and means of creating a microclimate, the
conservator should consider the following questions (Cassar 1984, 1985):
10
What are the requirements of the object, based on its environ-
mental history?
What is the climate in the gallery where the microclimate case
is to be placed?
What are the functions of the microclimate? Is it to act as a
stabilizing, dehumidifying, or humidifying factor to the object?
What will be the materials used for constructing the
display case?
11
The importance of using inorganic materials, such as glass and
metal, in constructing the case cannot be emphasized enough (Padeld,
Erhard, and Hopwood 1982). However, the buering material can be either
organic (wood, paper, textiles) or synthetic or natural derivatives (Nikka
pellets, Kaken Gel, zeolite clay, silica gel, Art-Sorb) (Weintraub 1982).
12
Thomsons recommendation of 20 kg of silica gel per cubic meter
for buering purposes in exhibition cases has been regarded as a good
starting point (Thomson 1977), but in certain circumstances, the same
result may be achieved with less. Recent research, however, questions the
recommendation of using any buering material at all in microclimate
boxes (Wadum et al. 1994).
Display materials also inuence the buering ability of a display
case and should therefore be chosen carefully. They should all be condi-
tioned before installation. Conditioning hygroscopic materials may require
up to one months exposure to the desired RH before the equilibrium
wished in the microclimate environment is achieved (Fig. 2ac).
Microclimate boxes with added buers
Even though most authors thought that wood itself could be used as a
buer, there was often a tendency to add an extra buer to stabilize the
internal RH of the microclimate box.
In 1933 a patent appeared for the use of salt-hydrate pairs as regu-
lating substances in cases and picture frames. The humidity should be con-
trolled through a low rate of air exchange, so that all the entering air passes
over certain salt-hydrate pairs. In this way, one salt may absorb moisture
from air that is too humid, while the other salt will conversely release mois-
ture if the air is too dry (Wilson and Barridge 1933). Shortly thereafter, in
Microclimate Boxes:
18921994
500 Wadum
1934, MacIntyre published test results to show that RH in a poorly sealed
display case is still more stable than the RH in the surrounding room. He
further demonstrated that the hygroscopic panel, frame, and fabric lining
of the case would improve this stability so that even with a 1 mm gap
around the glass base, a fairly constant RH could be maintained during the
week of monitoring (MacIntyre 1934). The results were applied to an air-
conditioning system for Mantegnas cartoons at Hampton Court Palace.
In 1934 Constable proposed an alternative to buers. The idea was
to feed conditioned air into the frame (or case) by means of pipes; how-
ever, this was dismissed at the time on the presumptions of bulk and
inconvenience (Constable 1934). The idea was nevertheless put into prac-
tice approximately fty years later (Lafontaine and Michalski 1984).
13
In 1936 Curister enclosed a panel painting attributed to Hugo van
der Goes. Salts were kept in trays within the base of the double-glazed
standing vitrine, which was capable of keeping a stable RH indenitely,
provided the exchange rate with the exterior was not too great.
14
Small
glazed openings were made at the top of the cases, through which
enclosed hygrometers could be monitored. Before the construction and
assembly of the microclimate box, the wood used in the construction of
the cases and frames was carefully seasoned and conditioned in an atmos-
phere of the agreed moisture content. During the most dicult climato-
logical months, the sealed cases showed a stable internal RH of 55%.
More than twenty-ve years would pass before a new description
of a microclimate box for a panel painting appeared (Sack 196364). Sack
describes how a controlled environment was made for a panel painting
and kept stable during a low winter RH of 1228%. A large sealed wooden
case with a double glass door was constructed that held pans containing a
saturated solution of magnesium nitrate hexahydrate. A small fan distrib-
uted the conditioned air to all areas within the case. In this manner, the
RH was held stable between 50% and 52%.
Shortly after, Stolow published his aforementioned studies of the
humidity and thermal properties of a sealed case (Stolow 1967).
15
If the elements (case and painting) are in equilibrium with the
environmental RH and temperature when the case is sealed and then
subsequently placed in another environment, a new equilibrium will
501 Mi crocii xart Boxts ror Pati Pai ri os
g p
a b c
Fi gure 2ac
Three main principles behind the construction
of a microclimate box: (a) a box containing a
panel painting and buer material; (b) a box
containing only a panel painting; and (c) a box
containing a panel painting and an altered
gaseous content.
a b c
develop within the case after a certain time.
16
Thus the sealed case
when tightly packed with conditioned wood and similar hygroscopic or
moisture-sensitive componentscan maintain reasonable RH control over
temperature changes.
There are two instances to which the above conditions do not
apply and where more complicated formulas must be used. The rst arises
if the case is not tightly packed; the second occurs when the internal air
volume is relatively large compared with that of the humidity-sensitive
materials. If the air volume is very large, the moisture properties of the
internal air dominate the relationship between RH and temperature; in
this case an increase of temperature will cause a decrease of the RH, and
vice versa. Stolow advises that silica gel be used to stabilize the RH, as the
response of the gel to temperature is negligible.
Based on the studies of Thomson and Stolow, Diamonds 1974
article on a micro-microclimate gave the rst description of a micro-
climate box for a panel painting on display. A sixteenth-century French
portrait from the school of Franois Clouet was placed in a showcase. It
appeared that with a maximum uctuation of temperature in the galleries
of 11 C, the RH should vary by less than 4%.
Accordingly, a hardwood box was constructed and tted at the
front with glass, which was puttied to make an airtight seal. A chipboard
back was made. This procedure yielded a box of approximately 13.7 l
volume, containing about 220 g of wood (picture and frame), which,
according to Thomsons gures, should have produced a near-stable envi-
ronment. The wood of the case was left uncoated so that it could play its
part in absorbing and giving o moisture. The whole box was conditioned
for two weeks to 55% RH (5%) and 20 C (2 C).
The fact that the picture showed signs of distress very soon after
being treated suggested either that it was sensitive to changes of RH of
less than 4% or that the design of the box was faulty.
The construction of a completely airtight box was impossible, due
to nances. Therefore, a buer was chosen to reduce the RH uctuations.
The principles involved were those laid out by Stolow (1966). The box was
tted with panels of silica gel held in a grid. The grid was crucial, as it
spread the silica gel over the largest area possible within the box.
17
The
open box and all its materials were left for four weeks to reach equilibrium
in a stable environment.
The environment was controlled with a small hygrometer and
was stable around 41% RH (4%) over two months. Variations inside the
box were no greater than 5%, so the box was considered a safe container
for the painting.
The box protected the painting from considerable uctuations of
approximately 20% during this period. Thus, only minor changes in RH
took place inside.
The same year Toishi and Miura described how the Mona Lisa from
the Louvre was exhibited for fty days in the Tokyo National Museum
(Toishi and Miura 1977). Throughout the run of that exhibition, the paint-
ing was enclosed in an iron case equipped with a double-panel glass win-
dow and lined with a 75 mm layer of glass. To maintain a stable RH of
50%, zeolite was placed in the case. The zeolite was found to be capable of
absorbing various gases such as sulfur dioxide, hydrogen sulde, ammonia,
carbon dioxide, and formaldehyde. The zeolite had been brought to a
humidity equilibrium in air at 60% RH (Kenjo and Toishi 1975).
502 Wadum
Probably the most-cited contribution on controlling microclimates
was written by Thomson in 1977. He derived a formula with experimental
support to predict the RH changes inside an unsealed exhibition case that
contained a buer such as silica gel. The formula showed that a well-
constructed case (containing about 20 kg silica gel per cubic meter of case
volume) should constrain seasonal humidity variation within reasonable
limits and, in some climates, make air-conditioning unnecessary. The prac-
tical solution recommended by Thomson was to make a showcase of non-
moisture-permeable materials and snugly tting closures, possibly gaskets.
For RH conditions above 50%, silica gel oers little advantage
over wood, as its M value is about the same.
18
However, at lower RH val-
ues silica gel is the best buer.
In this article Thomson does not take fully into account the
change of temperature; his focus is mainly on the RH changes. Tests of
the half-time of the case were made under constant temperature levels.
Also, the tests were conducted only with silica gel, not with other buer
materials, such as wood.
The leakage rate for the case is important. Thomson refers to
important studies by Padeld on the problem of diusion through various
materials (Padeld 1966).
19
Sack and Stolow (1978) reported that in a case designed in 1963 to
exhibit a German panel painting in the Brooklyn Museums main entrance
lobby (an area of the museum with a particularly erratic climate), a satu-
rated solution of magnesium nitrate hexahydrate proved to be eective in
controlling the RH at 5052%.
In another situation, a similar box served to control the micro-
climate around a painting on a thin wooden panel. This microclimate box
was constructed to protect a ne Fayum panel on loan to the Brooklyn
Museum. The intention was to design a case as airtight as possible to pre-
serve the required level of RH, independent of external variations. The
Fayum painting (44.5 28.5 0.2 cm thick) was painted on thin wood.
The wood had been bent to conform to the double convex contours of the
original mummy case.
20
It was decided to enclose the Fayum painting in a case kept at a
constant RH of 50%. Preconditioned silica gel would serve as the RH sta-
bilizing agent in the case. The case consisted of an outer display box and
an inner, airtight, metal-and-glass chamber. Inside the case, a wooden
frame was covered with fabric containing the preconditioned (50% RH)
silica gel, with the painting secured 4 mm in front of the silica gel panel.
A section of paper-strip RH indicator was placed in the corner of the case
to allow continuous monitoring of the internal RH. The painting attened
considerably from its convex warp while sealed inside this case.
Although the case was almost airtight, a very slow moisture
exchange with the exterior could still occur over time. This possibility
made it necessary to recondition the silica gel annually. Since it was time-
consuming to remove, recondition, and replace the silica gel, a second
panel was made. Kept under secure airtight conditions, it could be
installed as a replacement to the worn-out panel, which would be re-
conditioned and readied for the next annual replacement.
Acclimatization of two large (922 l) vitrines of air containing ve
icons was carried out to attempt the dicult task of stabilizing the gallery
environment at 5060% RH (Schweizer and Rinuy 1980). To keep the envi-
ronment stable, the recommended amount (20 kg m
3
) of silica gel was
503 Mi crocii xart Boxts ror Pati Pai ri os
placed in a honeycomb tray and covered with a nylon screen. With the
screen facing the interior, the tray formed the back of the case. The results
showed that the temperatures in the gallery and showcase were approxi-
mately the same at all times. In contrast, the RH within the cases
remained stable despite changes of 4474% in the RH outside the show-
cases. Evaluation of the amount of silica gel actually required to keep the
RH level stable in the vitrine led to a recommendation of 1015 kg m
3

almost half of what Thomson advised. It was also noted that the condi-
tioning of the silica gel should be at an RH value 5% higher than what was
actually desired in the case.
At the Sainsbury Centre for Visual Arts at Norwich, England, the
use of a mechanical system dependent on electricity was considered imprac-
tical to assess RH control employed within showcases (Brimblecombe and
Ramer 1983).
21
The use of a saturated salt solution, which is most eective
when auxiliary support is provided by an electric fan, presented the same
drawbacks as the fully mechanical system. The use of silica gel enabled the
creation of a self-sucient system without the need for electrical support.
To monitor the mechanism of air exchange between the interior
and the exterior of the case, an experiment was designed using a tracer-gas
method to monitor the concentration of various gases over time within a
standard-sized display case.
22
Padelds indication that the air-exchange
process occurs essentially by diusion was conrmed (Padeld 1966).
Additionally, Thomsons studies showing that the exchange of air within a
display caseand hence water-vapor variationoccurs exponentially were
also veried (Ramer 1981, 1985).
The conclusion reached, based on a calculation of the hygro-
metric half-time, was that Thomsons recommendation to use 20 kg m
3
of silica gel was valid.
The diusion of air is the primary cause of RH variation
within showcases; therefore, good construction of cases is essential
(Ramer 1981, 1985).
Also in 1981, a number of case histories about controlled-climate
cases were presented by Stolow (1981). One such case involved a large
panel painting and its predella by Neri di Bicci. The acrylic case enclosing
the panel was relatively small in air volume compared to the object vol-
ume, having only slightly larger dimensions than the artwork to allow for
maximum buering action of the silica gel. The estimated weight of the
panel and the predella was 250 kg. After consideration of the panel paint-
ing and the supporting materials (i.e., fabrics, wood), it was deemed neces-
sary to place inside the case approximately 200 kg of conditioned silica gel,
which was held in place by a screened panel covered with linen fabric.
With the past environment of the panel painting considered, it
was decided to establish a slightly higher-than-average RH (45%) within
the case. The EMC of the silica gel was periodically tested during the con-
ditioning procedure to verify, via sorption curves (isotherms), that the 45%
RH operating level had been reached.
Electronic probes were considered to monitor the interior of the
case, but because they are costly and require frequent calibration, they
were abandoned in favor of paper RH indicators. After one year of opera-
tion, it was shown that the internal RH level had been kept at a fairly con-
stant 4043% RH, despite wide variations in the gallery climate.
A further example of a specic microclimate box is to be found in
a description by Knight of the Tate panels in the Church of All Hallows
504 Wadum
Berkyngechirche by the Tower (Knight 1983). A box was made of Perspex
(known in the United States by the trade name Plexiglas), with a sheet of
aluminum as a backing board. Steel brackets attached the box to the wall,
thus leaving an air gap between the back plate and the wall.
Recommendations by Stolow and by Sack and Stolow provided
the basis for the humidity-control requirements of the box (Stolow 1977;
Sack and Stolow 1978). Silica gel was placed in the box in small narrow
trays that could be individually removed for reconditioning. After installa-
tion, a small hygrometer showed that the interior RH was maintained at a
level of 5658%.
The variation in RH in an experimental exhibition case that
was intentionally not sealed or airtight was monitored over two years
(Schweizer 1984). The RH of the surrounding room varied considerably
(2070%), but the RH inside the case, which contained silica gel, maintained
acceptable stability (4058%). This type of box, therefore, would prove very
useful in regions with hot summers and cold winters. The amount of silica
gel required was based on Thomsons formula of 20 kg m
3
.
Also in 1984, a microclimate box was presented by Ramer for a
seventeenth-century panel painting from the Netherlands (Ramer 1984).
The goal was to createwith a more aesthetic design than previous
microclimate boxesa humidity-controlled display case for the painting
that covered both the panel and frame. The new microclimate box was to
be tted into the extended rabbet of the picture frame, making this the
rst occurrence of its kind since the late nineteenth century (Simpson
1893) (see the section below entitled Microclimate boxes that alter the
gaseous content).
Practical requirements demanded a low maintenance level and
easy recharging of the silica gel humidity buer. The RH requirement
within the case was 55%. The silica gel amount was determined according
to Thomsons formula of 20 kg m
3
.
The microclimate box was made of inert materials (e.g., alu-
minum), and the glazing at the front was composed of 5 mm polycarbon-
ate sheeting (Lexan). As in previous designs, the tray of silica gel could
easily be remounted and reconditioned. The box was designed by
B. Hartley, A. Southall, and B. L. Ramer.
Thirteen Fayum mummy portraits and a panel painting of Saint
Luke by Simone Martini, all housed in the J. Paul Getty Museum in
Malibu, California, were placed in special cases that had a higher humidity
than normally maintained in the paintings galleries (Rothe and Metro
1985). An absolutely airtight microclimate box was constructed, with care
taken to make sure that it wasnt too visually overpowering.
23
The case
consisted of three basic sections: a back panel, a front bonnet (vitrine), and
a silica gel container. Art-Sorb was selected as the buer in accordance
with comparative performance statistics published by Weintraub and
Miura (Weintraub 1982; Miura 1981).
For the Simone Martini panel, 4 kg (dry weight) of Art-Sorb was
placed in the gel container and conditioned in a humidity chamber to 66%
RH. This amount is four times greater than recommended by Thomson
(1977) for a case of this size. The showcase had been on display since
March 1983 in a temperature- and RH-controlled gallery. The RH in the
gallery was always 1416% lower than the RH inside the case.
The same construction was used for the Fayum portraits, except for
the back panel, which was replaced by a Formica panel. The silica gel con-
505 Mi crocii xart Boxts ror Pati Pai ri os
tainer was made out of birch with a silk-screen fabric stretched over the
front and back. The gallery used for this display is open to the outside envi-
ronment during public hours, a factor that inuenced the RH, which ranged
from a low of 37% to a high of 68% during the test period. During the year,
the temperature ranged from 20 C to 27 C. The mummy portraits
required cases that were capable of maintaining an ideal environment of
50% RH, with minimal or no uctuations. After observation of the hygro-
meters in the cases, it was ascertained that the RH never varied more than
2%. Thus, it was not necessary to recondition the Art-Sorb for two years.
Because the cases were constructed of Plexiglas, the objects were clearly
visible and could be lit from the outside without any apparent change in
temperature.
Dissatisfaction with the microclimate boxes previously used by
the Kunsthistorisches Museum, Vienna, led Ranacher (1988) to present a
slightly dierent idea.
24
In his concept, silica gel could be renewed without
dismantling of the box, and an electronic device enabled convenient exter-
nal checking of the internal environment (Mayer 1988). The back and sides
of the box were made of wood to aid in stabilizing the internal moisture
content. The front of the box consisted of a Plexiglas hood, which was
mounted on the frame of the backing board. The frame of the painting on
display would be mounted over a hole in an internal wooden board cover-
ing the backing of silica gel. The amount of buer material (7 kg m
3
) was
determined by Ranachers own experimentation, not chosen according to
previously recommended high values of 1020 kg m
3
, or recommended
low values of 12 kg m
3
as recorded by Miura in his laboratory tests
(Miura 1981). The ratio used in Vienna had previously been proved ade-
quate for maintaining a stable RH of 50% within a microclimate box that
hung in a gallery having temperature uctuations of 1423 C. The built-in
electronic device for monitoring RH and temperature levels was invisible
to the public. Personnel could read the electronic data by plugging in a
wire at the bottom edge of the box.
At the United Kingdom Institute for Conservation conference,
Cassar and Edmunds individually presented microclimate boxes designed
to t within the frame of the painting, similar to those presented by
Ramer in 1984 (Cassar 1988; Edmunds 1988). Cassar enclosed a panel
painting in a buildup of the original frame, which permitted the manufac-
ture of a glazing (Perspex) and backing. The environment of the box was
kept at a stable RH through the presence of an Art-Sorb sheet placed
behind the painting. Edmunds constructed a closed box with low-
reection glass at the front and with Perspex sides and backing. A Perspex
grid containing conditioned silica gel crystals in small sacks could be
stored behind the panel painting. A hair hygrometer and, later, Grant
Squirrel Data Loggers were used to monitor the box interior and sur-
rounding environment. The data showed that the inside RH remained
stable for a considerable period at various ambient conditions without
recalibration of the silica gel. Cassar also reached the same conclusion.
Bosshard and Richard also recognized the disadvantages of micro-
climate boxes that enclosed both the painting and its frame (Bosshard and
Richard 1989). A box enclosing only the painting was developed and
widely distributed by Johnson and Wight in the beginning of the 1980s in
California.
25
This box was further rened, in conjunction with an empirical
trial with the Thyssen-Bornemisza Collection, to become a standard-climate
vitrine. This new microclimate box was at and could, therefore, be tted
506 Wadum
into the frame of the painting (Bosshard 1990). With low-reection glaz-
ing, the box could hardly be seen. The rabbet of the frame often had to
be extended to make room for the box, but in situations where this action
was not desirable, the sides of the vitrine could be made of a thinner
metal foil instead.
Art-Sorb granules were preferred to Art-Sorb sheets, as the gel is
more reactive in absorbing and desorbing moisture. The inside of the box
was made according to the specications: one-third panel, one-third silica
gel, and one-third air.
Because RH always drops after the box is closed, the Art-Sorb
was conditioned to a RH of 3% higher than desired. A paper RH meter was
placed in back, making it possible to check the RH inside the box at any
time. Foam rubber on the silica gel frame pressed the painting forward to
the front of the box. At present, more than fty-eight panel paintingson
loan or in the Thyssen-Bornemisza Collectionare kept in these vitrines.
Simultaneously with the empirical trial in the Thyssen-Bornemisza
Collection, Mervin Richard carried out lab tests at the National Gallery
in Washington (Richard 1993). The results showed that the thicker the
walls of the box, the greater its stability. The interior RH depends on the
amount of the buer material, and the greater the dierence between
RH outside and inside the case, the quicker the inside will change to a
new equilibrium.
Thomson recommended 20 kg m
3
of silica gel. As the Art-Sorb
in this case was deliberately over the requirements of the air volume,
overkill was established. Richard proved with his climate chamber that
a temperature change of 10 C resulted in a change of about 2% RH
inside the box, depending on its size and capacity to absorb the tempera-
ture change.
26
In 1990 a microclimate box to be tted within a frame was con-
structed in the Mauritshuis, The Hague, largely following the concepts of
Ramer, Bosshard, and Edmunds (Wadum 1992).
27
The glazing was, how-
ever, always a layered safety glass that enabled the box to travel with mini-
mum risk.
28
At rst the box included silica gel or Art-Sorb sheets to
stabilize its internal RH during display and transit (Wadum 1993).
29
Between the glazing and the front of the painting, in the rabbet, a grid
was placed along all four sides allowing convection of the air from front
to back and vice versa.
Small built-in microprocessor loggers monitored the RH and tem-
perature from the time of installation until the painting was returned after
loan.
30
The printout showed that the RH stayed stable within 2%, despite
temperature uctuations of more than 10 C.
Simultaneously with the Mauritshuis, the Rijksmuseum in
Amsterdam was also developing a microclimate box. This box, a low-
budget variant, was initiated and constructed by Sozzani, who needed a
simple, easy-to-mount box to t into the frame (Sozzani 1992). The box was
constructed of safety glass that was mounted and sealed in the rabbet of
the frame. Behind this, the painting was mounted in the usual way. Thin
wooden battens were built up on the back of the frame, allowing enough
depth in the rabbet for the insertion of a sheet of Art-Sorb behind the
panel. The stainless steel backing sealed o the box with airtight gaskets.
The primary advantage of this type of box is that the rabbet never
has to be extended, a requirement that would be undesirable in many situa-
tions. The previously used microclimate boxes from California required
507 Mi crocii xart Boxts ror Pati Pai ri os
some manipulation of the frame.
31
The Rijksmuseum boxes also proved
eective when monitored with humidity indicator strips or small hygrome-
ters, all of which indicated a stable RH within the boxes in the museum
environment.
Extensive studies undertaken by Richard have conrmed that tem-
perature changes aect panel paintings much faster than do RH variations
(Richard 1994). Although he concludes that silica gel has no eect on the
temperature changes, he nevertheless recommends that the gel remain in
use for microclimate boxes. Drawing on the assumption that virtually all
microclimate boxes leak, Richard states that silica gel plays an important
role in stabilizing the RH in display cases used in unsuitable environments
for extended periods.
Microclimate boxes without added buers
A more recent approach to the construction of microclimate boxes relied
on the hygroscopic behavior of the wood panel itself as a stabilizing factor
within a small volume of air. Such boxes were not kept at a stable RH
through added buers but instead maintained their own internal moisture
equilibrium at changing temperatures.
A critical approach to the consistently recommended use of a
moisture buer in small display cases was presented by Ashley-Smith and
Moncrie (1984). Their experiences in the Victoria and Albert Museum in
London showed that the silica gel in a showcase neutralizes the short-term
RH uctuations but does not compensate for seasonal changes. Ashley-
Smith and Moncrie concluded that for wooden showcases, silica gel gives
poor results in relation to the time and expense required to purchase, pre-
pare, and handle it, as well as to design and build showcases to accommo-
date it. They stated that an ordinary showcase without silica gel fares
nearly as wellor as poorlyin reducing short-range uctuations. The
same conclusions were drawn in reference to some old-fashioned walnut
cases in the Royal Ontario Museum, Toronto, that proved remarkably
eective in slowing moderate uctuations of RH (Phillimore 1979). For
best results, a well-sealed case made completely of metal and glass or plas-
tic is usually essential (Brimblecombe and Ramer 1983). However, for the
Victoria and Albert Museum, wooden case vitrines serve in themselves as
useful, additional buers (see Cassar and Martin 1994).
Also in the early 1980s a special type of microclimate box was
created by Padeld, Burke, and Erhard (1984). A cool-temperature display
case was made for a vellum document placed in a close-tting airtight con-
tainer. The document required a stable temperature of 16 C, some six
degrees cooler than the gallery, and an RH of 4050%. The box maintained
a nearly constant RH after cooling; however, special care was necessary to
minimize temperature gradients. The case performed satisfactorily for one
year with no change in internal moisture content.
The simplest method possible was chosen for displaying this
document. It was sealed inside a thin, airtight container that was cooled
by means of the Peltier eect.
32
The refrigeration system of the box con-
sisted of two coolers at the bottom of the aluminum tray holding the
microclimate box.
A close-tting, airtight enclosure has many advantages for the
temporary exhibition of at pieces of vellum or paper. It can be designed
to maintain a nearly constant moisture content and a safe RH. At room
508 Wadum
temperature, paper contains thousands of times more water than an
equal volume of air does. In a sealed box full of paper, therefore, it is
the paper that controls the RH of the surrounding air, if both are of the
same temperature.
Based on the psychrometric chart, it was obvious that a container
holding more than 1 g of paper per liter of air has a reasonably stable RH as
the temperature varies (a rule of thumb that, incidentally, holds true over
the whole range of ambient temperature). This conclusion applies only to
a slow temperature change imposed uniformly to the paper and box.
It is important to remember that absorbent material such as
paper or silica gel only functions as an RH buer if it is at the same tem-
perature as the air or object to be buered. To buer for eventual air leak-
age of the sealed box, extra paper was enclosed in the box to increase the
buering capacity.
Apart from using inert material for the inside of the box, a further
precaution against air pollution involved using paper containing calcium
carbonate to absorb acid gases.
In 1987 Hackney warned against enclosing buering materials
such as silica gel in small, sealed environments. He underlined, as have
authors before him, that the equilibrium of silica gel or similar buers is
not dependent on changes in temperature (Stolow 1965, 1967; Thomson
1964, 1977; Weintraub 1982). On the contrary, hygroscopic materials such
as wood were characterized by relative equilibrium, showing a higher RH
at higher temperatures, and vice versa.
Despite these developments, the creation of microclimate boxes
continued with added buers such as silica gel or Art-Sorb (as discussed
above in the section entitled Microclimate boxes with added buers).
The tradition continued, under the inuences of guidelines laid out by
the authors mentioned above, to keep the internal RH stable under all
circumstances.
Richard reported in 1991 that in closed cases, falling RH levels
caused by temperature decreases should not cause alarm, noting that
several publications have emphasized that it is not benecial to maintain
stable RH levels for hygroscopic works in transport if temperature changes
are anticipated at the new location. If, for example, a painting were moved
from 50% RH and 20 C into a very cold gallery, a lower RH must be
maintained if the EMC is to be kept constant within the object.
Users of microclimate boxes seemed fairly reassured by the stable
RH values produced through the use of added buers such as silica gel or
Art-Sorb. However, considerations regarding the eects of temperature
uctuations on the wood of the enclosed panel developed into an exten-
sive test program set up by the Mauritshuis, The Hague; the Central
Research Laboratory for Objects of Art and Science (CL), Amsterdam;
and the Rijksmuseum, Amsterdam (Wadum et al. 1994).
The tests at the CL demonstrated that buering material should be
avoided in small microclimate boxes. Otherwise, uctuations in the tem-
perature would initiate a breathing process between the non-temperature-
reactive silica gel or Art-Sorb and the panel.
Boxes made of inert material proved eective in maintaining
stable environments for the hygroscopic material inside. A box made of an
inert front and back, but placed in the wooden rabbet of the frame, also
provided eective maintenance against uctuations of 1030 C. Long-
term (i.e., more than eight hours) low or high temperatures were not
509 Mi crocii xart Boxts ror Pati Pai ri os
tested. RH uctuated between 30% and 70% without any inuence on
the interior climate. The boxes were well sealed to prevent leakage.
The Mauritshuis microclimate box now uses polycarbonate sheets
as a backing; because buer material is not used, the reverse of the paint-
ing is left visible so that the courier or other museum sta can examine it
without removing it from the microclimate box.
33
Dimensional movement of dierent types of wood in closed
cases, with and without silica gel, was studied by Kamba (1993). He states
that the dimensional change of the wood inside the box without silica gel
was less pronounced than that of the wood in the silica gelbuered case.
Kambas studies thus conrmed the results from the tests at the CL, in
which an equilibrium between wood and the surrounding air at dierent
temperatures was attained without added buers.
For these reasons the most recent microclimate boxes for panel
paintings at the Mauritshuis and the Rijksmuseum are now made without
any added sorbent material. The buering role of the panel itself is
regarded as sucient for the small, enclosed environment of a microcli-
mate box. However, care is taken to ensure stable temperatures around the
microclimate box, whether it is on display in the gallery or in transport
(Wadum et al. 1994). To this end, the research at the CL also showed that
maintaining an open air space of 2 cm or more between the microclimate
box and the wall increases considerably the stability of temperature within
the box (see also Ranacher 1994). Thermally insulated transit crates may
maintain a relatively stable temperature inside the microclimate box on
long journeys (Fig. 3ad).
Microclimate boxes that alter the gaseous content
Apart from one very early foray, the use of microclimate boxes with
an altered gaseous content has become popular only in the last decade.
This new interest arose from the need to reduce the deteriorating
eects of oxygen.
The rst known attempt to make a microclimate box was in 1892 in
England by Simpson, to protect a painting by J. M. W. Turner in the Victoria
and Albert Museum (Simpson 1893). The characteristicstailored to t the
specic paintingof this sealed, airtight box were very similar to a modern
microclimate box. Simpsons box was even intended to be tted into the
original gilt frame and hung in the usual manner. The front was composed
of glass; the back comprised glass, metal, or other materials. In Simpsons
box, nozzles were placed at the bottom for attachment to an exhauster,
which could extract air from the box to create a vacuum around the picture.
510 Wadum
30 Wadum fig 04 eps
a b c d Fi gure 3ad
Four main types of microclimate boxes: (a) a
box containing a panel and buer and no
framing, (b) a box encapsulating a framed
panel and buer, (c) a framed box containing
a panel and buer, and (d) a framed box con-
taining only a panel.
Simpson concludes his description by asserting that the color of
the picture in the box would be hitherto immune to light, sun rays, damp-
ness, or other damaging external inuences.
34
Indeed, time has shown that
the Turner painting is in excellent condition to this day; until the present,
the box has not been opened. Although hardly subject to vacuum for very
long, Simpsons box represents the rst attempt to create an altered
gaseous content around the object enclosed in the microenvironment.
The rst inert gas display case was described by Byrne (1984). An
egy gure from Easter Island was placed in a round Plexiglas tube acting
as a display case. The ends were sealed with Plexiglas disks tted to the
tube. Silicone rubber served as a gasket. The tube was 20 cm in diameter;
its walls were 6.3 mm thick. To avoid the presence of water vapor around
the egy gure, the tube was charged with nitrogen gas to exclude oxy-
gen and moisture. A modied aneroid barometer monitored the pressure
within the case and conrmed the presence of a stable charge of nitrogen
gas. Four years later the case showed a loss of pressure, so nitrogen gas
was added again. A humidity indicator strip was placed in the case, and
future recharging with nitrogen was accomplished by rst bubbling the
gas through a water bath.
The use of Ageless as a means of generating low-oxygen atmos-
pheres for the treatment of insect-infested museum objects is discussed
by Gilberg (1990). Ageless is a type of oxygen scavenger that is described
by the manufacturer to be a mixture of nely divided moist iron, (ferrous)
oxide, and potassium chloride, a combination that rapidly absorbs atmos-
pheric oxygen. The oxygen concentration in a microclimate box can be
reduced to less than 0.05% as the introduced Ageless quickly reacts with
any oxygen leaks. Ageless can also reduce the oxygen concentration in a
closed environment to less than 0.01% and can maintain this level
indenitely, depending on the permeability of the packing material.
Ageless is available in dierent package sizes that correspond to
the amount of oxygen to be scavenged (for example, Ageless Z-200 is
capable of absorbing the 200 ml of oxygen contained in 1 l of air). Ageless-
Eye is an oxygen indicator in tablet form that changes color in relation to
the absence or presence of oxygen. Tests in which insect-infested objects
were kept at 30 C and 60% RH resulted in convincingly stable, low oxy-
gen levels and stable RH.
Ageless is being used to prevent deterioration of rubber, which
becomes brittle as a result of ultraviolet light, ozone, and oxygen
(Shashoua and Thomson 1991). After some rubber objects in the British
Museum, London, were sealed in bags, the oxygen was reduced; an investi-
gation into the deterioration rate of the objects showed positive results.
Further investigations on the uses and reactions of Ageless were
undertaken at the Getty Conservation Institute to develop hermetically
sealed, inert, gas-lled display and storage cases (Lambert, Daniel, and
Preusser 1992).
No matter how well cases are designed and constructed, some
air can always enter. If their value as oxygen-free chambers is to continue,
the leaking cases must be reushed with nitrogen or some other inert gas.
After the original ush, the oxygen-free life span of the case can be greatly
extended by an oxygen scavenger placed in the case. Calculation of the
approximate lifetime of a case is obtained by dividing the oxygen-absorbing
capacity of Ageless in the case by the leak rate per day.
511 Mi crocii xart Boxts ror Pati Pai ri os
The Getty Conservation Institute studies were conducted on
packets of Ageless-Z in boxes, in which RH-conditioned nitrogen was pro-
duced by control of the mixing ratio of dry nitrogen obtained from the
cylinder, to humidied nitrogenthe result of dry nitrogen bubbling
through water at room temperature (Byrne 1984). The test chamber was
initially ushed with nitrogen until the oxygen reached the 10009000 ppm
range. At this point Ageless was rapidly inserted and the test chamber her-
metically sealed. The RH inside the chamber was maintained at 52% with
saturated salt solutions (magnesium nitrate). This research showed that
Ageless reacts rapidly and thoroughly with oxygen in a sealed case that is
lled with an inert gas, and that has an optimal RH above about 50%.
Sealed cases lled with inert gas prevent the oxidation of the
objects placed therein. In small exible containers with little air content,
Ageless can perform well in spite of slight warming. It is hazardous, how-
ever, to place Ageless in a large rigid case containing air because of the
heat produced and also because of the risk of implosion when the oxygen
(20% of air) is removed. A sealed case lled with an inert gas should have
exible bellows attached, to compensate for temperature and pressure
uctuations in the museum atmosphere.
A slight color change in cinnabar, litharge, and sienna has been
observed on objects in nitrogen-lled sealed cases (Toshiko 1980). There
is good evidence, however, that a nitrogen atmosphere retards the fading
of watercolors.
The Getty Conservation Institute, as well as Gilberg and Grattan,
concluded that Ageless is a rapid and ecient oxygen scavenger (Gilberg
and Grattan 1994). Its use in an inert, gas-lled, hermetically sealed display
case with a moderate leak rate should maintain the oxygen content at a
very low level for several years. An environment with an RH of 53% or
above is recommended. Both the level of the oxygen content and the inter-
val after which an Ageless-equipped case will require a replacement and
ushing can be readily predicted if the case leakage rate is known.
There are many devices for measuring RH; they range from aspiration
and sling hygrometers to thermohygrographs, dial hygrometers, cobalt
salt strips, and data loggers of various kinds. Thomson and Brown have
described the pros and cons for a number of devices, showing how unreli-
able they can often be, either because of an instruments poor accuracy or
lack of calibration or because of mistakes made by the person manipulat-
ing the instrument (Thomson 1981; Brown 1994). Suggestions for the
monitoring of showcases include a special built-in sensor with digital read-
out or a printer (Mayer 1988). A number of small measuring devices have
also been used to keep track of activity inside the microclimate boxes.
Diamond placed a small Edney dial hygrometer inside the box,
after checking it for accuracy against a sling psychrometer. Diamonds
microclimate box covered both picture and frame, so the hygrometer
could be placed at at the bottom of the vitrine, enabling the viewer to
monitor the environment from the front of the box (Diamond 1974).
The vitrines used by Rothe and Metro of the J. Paul Getty
Museum had been tested with small thermohygrographs from Pastorelli
and Rapkin (Rothe and Metro 1985).
35
They were not as accurate as much
larger and more sophisticated thermohygrographs but were, in this
Measuring Devices
512 Wadum
instance, proved to be reliable, since they provided warning about air leak-
age. According to Rothe and Metro, the only evident disadvantage is a
necessity for frequent monitoring because no printout (that can be read
later) is produced.
Paper RH indicators with impregnated bands of cobalt salts
change from pink to blue in relation to the ambient RH. This type of indi-
cator has been used by most modern authors, and a thorough investigation
into their eectiveness has indeed proved them to be reliable and long last-
ing (Daniels and Wilthew 1983). A reference color against which to com-
pare the RH values on the strips is recommended.
36
As dial RH measuring
instruments have hair, paper, or special plastic sensing elements, they need
frequent recalibrating; strips, in contrast, are not altered over time.
Placement of the cobalt strips next to the painting within the vit-
rine is necessary to obtain an accurate reading. Since this is aesthetically
not a very pleasant solution and distracting for spectators, other place-
ments have been explored. The cards have often been placed on the back
of the boxes, but microclimate boxes that t within the frame can only be
monitored when the painting is turned, a procedure that requires much
time-consuming and unnecessary handling of the object in order to track
the changes in the microclimate box.
When daily monitoring of a microclimate box and its painting is
not feasible, a continuous record of activity is possible only with small
data loggers. Inspired by the National Gallery in Washington, D.C., the
Mauritshuis began monitoring the RH and temperature within microcli-
mate boxes using ACR data loggers (Wadum 1992).
37
The small logger was
mounted behind the panel on the inside of the backing lid of the microcli-
mate box, with its communication socket in the frame of the vitrine. This
method allowed for initialization of the logger inside the box without its
being opened. When the painting was traveling, the courier made backups
of the logged RH and temperature after arrival at the destination museum.
38
Then, a new interval of logging (typically around three months) was set
for the loan period to follow.
39
The courier and the registrar could then
evaluate the transit period and eventually arrange for improvements before
the return of the painting. These small loggers make it possible to keep a
complete record of a specic paintings climatological history, starting
from the moment of installation.
40
Discussing the aesthetics of microclimate boxes can initiate a heated
dialogue between most curators and conservators, as well as among the
public. Most people would probably prefer being close to an object of
study, without having the feeling of looking into a vitrine. Paintings in
vitrines seem remotethe vitrine forms a barrier between the spectator
and the artwork.
As previously discussed, microclimate boxes have developed from
vitrines hanging on the wall, enclosing painting and frame inside, to small
boxes placed behind and within the frame. This evolution clearly reects
the goal of distracting the spectator as minimally as possible. De Guichen
and Kabaoglu once made an ironic list of recommendations regarding the
optimum manufacture of a showcase (de Guichen and Kabaoglu 1985).
Almost all of their guidelines could also apply to the microclimate boxes
(to wit: one suggestion, to be sure to display the locking mechanism
Aesthetics
513 Mi crocii xart Boxts ror Pati Pai ri os
prominently, reects the assembling screws or painted backing boards
that make a disturbing impression on many a microclimate box).
During the installation of a painting in a microclimate box, dust
can become a nerve-racking nuisance (Avoid sealing the showcase too
tightly, because exhibits always look better when covered with a uniform
coat of dust, de Guichen notes). Many microclimate boxes on display do
show small specks of dust on the inside of the glass, and cleaning them
out is impossible without dismantling the whole box, a practice usually
acceptable only when the box has returned to the controlled environment
of the lending museum.
The protective Perspex or glass is another main issue. Many micro-
climate boxes recall de Guichens helpful suggestion to polish the glass of
your showcase to a mirror nish. Any glazed painting, particularly a darker
one, reects at certain viewing positions. Perspex has the most reective
qualities; coated and low-reection glass can reduce the amount of reec-
tion to a minimum. In some instances, detection of the protective glass in
front is impossible without specic inspection (Saunders and Reeve 1993).
The small (366 257 mm) Franois Clouet picture that Diamond
placed in his microclimate box is aesthetically and physically delicate
(Diamond 1974). It has an extremely nely wrought rosewood frame inlaid
with silver and mother-of-pearl, clearly not the sort of thing you just put
in a box and screw to the wall, he states. The proportions of the box, as
well as the color and texture of its lining, were thus critically considered
in the design; ultimately, the museum agreed that the picture actually
beneted from its more aesthetic installation, as well as its new, larger
presence on the gallery wall.
This particular approach for a small picture has also been used in
the display of fragments of altarpieces on gallery walls. These so-called
shadow boxes not only serve as buers but also enhance the objects physi-
cal presence.
Rothe and Metro state that microclimate boxes should not be too
visually overpowering, since their main function is to protect the painting
(Rothe and Metro 1985). Rothe and Metros Perspex box for the Simone
Martini also covered the paintings original, inseparable frame; the box
around the Fayumswhich, for obvious reasons, do not have frames
could, of course, only be of a showcase type. Here the objects became,
in a sense, archaeological fragments; without the microclimate boxes, the
visitor would not have the opportunity to view these fragile objects.
With a microclimate box covering both the painting and the frame,
the vitrine does not have to be built to t the panel painting exactly. Rather,
it can be made in standard sizes, allowing reuse for another painting at a
later date. Disadvantages include the high reection factor of Plexiglas and
the fact that some viewers nd the box aesthetically displeasing.
Ramer, however, suggested that to fulll aesthetic requirements,
the microclimate box around his Netherlandish painting should be tted
into the extended rabbet of the picture frame (Ramer 1984). The box in this
case pretends not to be present, leaving the viewers attention focused on
the painting. Most of the more recent constructors of microclimate boxes
(i.e., Cassar, Edmunds, Bosshard, Wadum, Sozzani) included these consider-
ations, preferring small, narrow boxes made to t behind the frame.
The use of low-reection glass of low iron content (which takes
the green out of normal glass) has limited the amount of disturbance
to a minimum.
514 Wadum
Encapsulating panel paintings in microclimate boxes in this manner rein-
forces the protection and care of our cultural heritage, benets that pro-
mote an increased willingness by museums to lend their most vulnerable
panel paintings.
It would be wrong, however, to suggest that all problems can be
overcome by tting a panel painting in a microclimate box. More secure
microclimate boxes with better seals against leakage have yet to be made.
Also, the problem of adequate thermal buers when a painting is on loan
has not, in many instances, been satisfactorily handled. The level of shock
or vibration to which a paint lm and its carrier are exposed during transit
still begs further denition: a better solution to this trauma must be found.
Correct acclimatization in historic buildings and museums also requires
much more research and attention, if the dimensional movement of
painted wood that is displayed or stored is to be stabilized.
The author is grateful for help and suggestions from Nicola and Nic
Costaras and from Feroza Verberne. Special thanks are oered to
Aleth Lorne and Victor Wadum for their support during the preparation
of this article.
1 Standards for sealed transport cases of wood painted with water-resistant paint, or lined on
the inside with a nonpermeable water-resistant membrane, are given by Stolow (1965). The
standards include precise volumes for wood and silica gel in the cases.
For maximum thermal insulation, a case should have thick walls, high thermal capacity,
small thermal conductivity, and small surface area (Stolow 1967). Stolow gives examples from
air transit, in which hulls of planes may reach temperatures of 40 C, or in which hulls have
no pressure correction, and therefore, at low pressure, air escapes the box. Upon a planes
return to earth, air again enters because of the higher pressure, and this air may be of an
undesired climatological condition. Therefore, cabin-pressure control and temperature control
during air transit are important factors to take into account.
2 Buck concludes that while good moisture barriers may almost completely insulate a panel
from short-cycle humidity variations, they may nonetheless be surprisingly ineective
against seasonal cycles. For recent studies on moisture buers applied on panel paintings,
see Brewer 1991.
3 Buck suggests that the larger uctuations in RH in the United States could be the reason for a
tendency to cradle panels more often in the United States than elsewhere (Buck 1962). He fur-
ther demonstrates that a cradled test panel that was kept in a heated, dry room for several
months showed shrinkage of roughly 1.4% in its width, with the members of the cradle stick-
ing out at the sides. Buck invites rheologists to communicate with restorers to learn about the
laws that govern the ow and deformation of materials.
4 The addition of hygroscopic material (having the same quick response as gelatin) at the rear of
the canvas and the sealing of the reverse by a loose lining would help reduce the rate of
response of the glue. Glazing with acrylic and a backboard creates further enclosure for the
original object and thus provides protection from unwanted reactions to temperature changes
(see Hackney 1990).
5 Investigation of thermal properties of transport cases is important when traveling exhibitions
are on the move. During travel, the cases may be exposed to unforeseen temperature condi-
tions, and the use of thermal linings can oer signicant protection and permit greater RH
stability within the cases (Stolow 1966).
It is also possible to maintain constant moisture content of soft-packed paintings by con-
trolling temperature, provided that the moisture barrier used as a wrapping material (polyeth-
ylene) is well sealed (Saunders, Sitwell, and Staniforth 1991).
An early example of polyethylene as a tight wrap for paintings coming from Europe to
Canada is recorded by Thomson (1961).
Notes
Acknowledgments
Conclusion
515 Mi crocii xart Boxts ror Pati Pai ri os
6 When wood and other moisture-containing materials are heated, they give o moisture. At
the same time, heated air can hold more moisture; so together the wood and the air reach a
new equilibrium. In an empty case of nonabsorbent material such as glass or metal, a rise in
temperature will cause a fall of RH, and vice versa. In a case holding a quantity of wood, the
situation is reversed: a rise in temperature will cause a rise in RH. When wood gets hotter, it
will give up moisture unless the surrounding RH rises. In a closed case, the RH will indeed
rise because of the moisture given o by the wood, and the two tendencies will counteract
each other. At median humidity, wood contains about twelve times as much moisture as air,
volume for volume. Therefore, wood or other cellulosic materials will have the dominant
eect on the interior of a small microclimate box.
Thomson showed in practical experiments that a ratio of 120 g wood per 100 l air achieves
a constant RH at changing temperatures (Thomson 1964). The change of RH will not exceed
about one-third of the temperature change (C) and will be in the same directionprovided
that there is no entry of outside air of a dierent RH into the case. For ratios greater than
1 kg of wood to 100 l of air, the standard curves for wood equilibrium may be used.
7 Based on the rather dramatic climatological changes occurring in Canada, Stolow demonstrates
his ndings on dierent forms of small environments within packing cases (Stolow 1967). It is
seen that a sealed case is capable of maintaining a certain level of RH when it contains wood or
similar cellulosic materials preconditioned to the desired level. The use of silica gel permits
exposure to even greater external temperature changes while it retains the same RH control.
8 The diusion coecient of water vapor through air is about 0.24 cm
2
sec
1
(Padeld 1966).
This is about twice the coecient of the other gases found in air. The coecient for diusion
through wood is about 1.2 10
4
cm
2
/sec for water vapor, and 0.75 10
4
cm
2
/sec for car-
bon dioxide (see Stamm 1964). This means that 1 m
2
of wood allows as much air to diuse as
3 cm
2
of hole through it, and it leaks water vapor as fast as a 5 cm
2
hole.
9 Weintraub introduces a number of tools for determining which sorbents will be most ecient
within a specic RH range (Weintraub 1981). In the 1978 International Council of Museums
Conference on Climatology in Museums, there was a general consensus that a sorbent should be
temperature independent and have as large a surface area as possible (e.g., powdered silica gel).
10 As a consequence of the many dierent types of microclimate vitrines being introduced by
various authors, Cassar proposed standardization of symbols to be used in classifying the
more commonly used types of case construction designs (Cassar 1984).
11 Many woods (especially British and European oak) give o organic acid vapors, which can
accumulate and harm many types of objects, including those of metal, marble, materials such
as mother-of-pearl and shell, and paper and textiles, in cases where the exchange of air
between inside and outside has been reduced to a minimum. All adhesives, adhesive tapes, and
sealants used should be tested for stability to ensure that none give o harmful vapors.
12 The choice of the right sorbent is essential and should be considered together with the RH
level required for the specic object. Therefore, it is essential to consider the isotherms for the
dierent kinds of sorbents before a decision is made.
13 The RH-control module designed to service a number of display cases is based on a mechani-
cal system combined with a buering agent such as silica gel (Lafontaine and Michalski 1984).
A plastic tubing system distributes the well-conditioned air to a number of display cases, rely-
ing on an air exchange in the display cases of a certain amount per day. Air in the display
cases equipped with this humidity-control module should be supplied at a rate of at least
double the natural leakage. One RH-control module can thus control many display cases. The
conditioned air enters the cases through the tubes and leaves again via natural openings that
permit leakage. There is no active temperature controlthe module passively follows the
room temperature. The system, therefore, works only if none of the cases is cooler than
the control module.
14 The salts used were hepta- and hexahydrates of zinc sulfate, which are at equilibrium in an
atmosphere of 55% RH at a temperature of 15 C (Curister 1936).
15 Stolow gives as an example a case for which the wood and silica gel are both 1000 g and the
RH is kept stable (Stolow 1967). Even a smaller ratio of gel to wood would have a stabilizing
eect, buering the internal RH against temperature changes. If silica gel is used, it should be
packed in a way that gives it as large a surface area as possible.
516 Wadum
16 The change of RH is somewhat more than a third of the imposed temperature change, and in
the same direction as the change (e.g., if the initial RH were 50%, the temperature 20 C, and
the case exposed to 30 C, the resulting RH would be 53.5% RH; if the case were exposed to a
temperature of 10 C, the nal RH would then go to 46.5%).
17 Stolow recommends a silica gel granola, not exceeding 3 mm, spread out thinly over as
large a surface as possible. He also advises the use of a dry weight of silica gel at least double
the weight of the material to be protected (Stolow 1966). In the box discussed, 450 g of silica
gel was used.
18 The M value is the specic moisture reservoir (moisture gain in g/kg for a 1% rise in RH).
19 Theoretical and experimental research at the Canadian Conservation Institute has shown that
if gaps at the top and bottom seams of a case are smaller than 0.3 mm, the leakage rate of the
case will be less than two air changes per day (Michalski 1985).
20 Previously the panel underwent conservation treatment as follows: The reverse was covered
with Saran F-300 (a copolymer of vinylidene and acrylonitrile, soluble in methyl ethyl ketone)
and a layer of glass fabric, in an eort to stabilize the panel. Prior to this treatment, it was
noted that there was a dark, water-soluble layer (skin-glue sizing perhaps) between the paint
lm and the wooden support. Four other Fayum portraits (two painted in encaustic, two in
a water-soluble medium) were examined, and it was concluded that the intermediate layer
between paint and wood was indeed very hygroscopic. The Saran and glass fabric on the
reverse side of the Fayum on loan may have altered the warpage pattern, as the panel devel-
oped a pronounced concave conguration.
21 The museum display cases used in the Sainsbury Centre and their exchange of water vapors
are being evaluated. The hygrometric half-time is calculated, as is the half-life for water
diusion in the cases. The better sealed the case, the longer the half-life (Brimblecombe
and Ramer 1983).
22 A large amount of nitrogen was passed into the case, via the screw hole where the Perspex
top was secured. Increasing the concentration of nitrogen acted to deplete the oxygen level
to approximately half its normal value. Immediately after the introduction of the nitrogen, a
small volume of carbon dioxide was added, which increased the carbon dioxide level of the air
in the case to about ten times its normal value. The following day, small samples of gas were
extracted and injected into a gas-liquid chromatograph in order that the oxygen and carbon
dioxide content might be determined. In this way the gradual loss of carbon dioxide and the
invasion of oxygen could be monitored. The half-lives for the exchange of oxygen and carbon
dioxide gases with the display case were calculated to be 2.3 and 2.7 days, respectively.
23 The case was designed in collaboration with Helmuth Guenschel, Inc., Baltimore, which actu-
ally built the case.
24 Ranachers concept was based on the microclimate boxes from the Philadelphia Museum of
Art (Ranacher 1988).
25 This box was made by the California company of G. F. Wight Conservation, following the
principle laid out by Bosshard and Richard.
26 This result is explained by the specic characteristics of Art-Sorb, which according to Bosshard
desorbs or absorbs dierent amounts of humidity depending on temperature. However, con-
tradictory reports by several authors as to the nature of the silica gel or Art-Sorb emphasize its
stability despite changes in temperature (Richard 1991).
Richard tested two vitrines of dierent size: one with an RH of 50%, the other with an RH
of 30%. After three months the RH in the small vitrine had decreased to 1%, the large vitrine
to only 0.5%. This result proves that the half-time will be around two years for the less sealed
of the two. Both tests were made in empty vitrines. It is concluded that the climate would have
been even better with the panel inside, as the hygroscopic material would help stabilize the
microenvironment. During transit the same benet was recorded: 16 C uctuations in the
vehicle but only 2 C uctuations in the box. RH uctuations of 45% were recorded in the
vehicle, but only 1% were recorded in the box, as it was kept in a well-insulated transport crate.
27 The box was made as a joint project with the Museum Boymansvan Beuningen, Rotterdam,
which had the skilled technical sta required for its production. Nicola Costaras, Luuk Struik
517 Mi crocii xart Boxts ror Pati Pai ri os
van der Loe, and Carol Pottasch all contributed to creating this rst box, which was designed
by Andr van Lier (Wadum 1992).
28 The safety glass used for the rst model was Noviex; at present the thinner and less-costly
Mirogard Protect Magic, Low-Iron, is used.
29 A method later found not advisable (Wadum 1993).
30 ACR data loggers from ACR Systems, Inc., were used. They were typically set at measuring
intervals of 30 seconds during transit and at 10 minutes throughout the duration of the loan.
31 See note 25 above.
32 The Peltier eect describes the absorption or emission of heat when an electric current passes
across the junction of two dissimilar conductors.
33 The microclimate boxes were initially made by Smit Mobile Equipment B.V., Oud-Beijerland,
the Netherlands; they are now produced by the technical sta of the museum, according to
the most recent manual.
34 The author is indebted to Susannah Edmunds at the Victoria and Albert Museum for informa-
tion on this early microclimate box.
35 Pastorelli and Rapkin Ltd., London, was taken over in 1983 by M and T Precision Instruments
Ltd., Eneld.
36 The Humidical Corp. type card no. 6203-BB seemed to satisfy most users.
37 The author is indebted to Sarah Fisher, National Gallery of Art, Washington, D.C., for sharing
her information on measuring devices.
38 Shock monitoring may also constitute part of the recording of a painting in transit. The most
recent literature on this topic can be found in Mecklenburg 1991, in which several authors deal
with the subject. The author has had fruitful discussions on this topic with David Saunders of
the National Gallery, London.
39 The logging interval during transit would often be 30 seconds; the interval during exhibition
would generally be 10 minutes.
40 With regard to the investigation into the performance of humidity sensors, M. Cassar is con-
ducting a comparison of ten dierent sensors for stability, drift, and long-term performance.
This work in progress will provide valuable information for the assessment of measurements
obtained by study of artifacts on display or in transit.
ACR data loggers, ACR Systems Inc., 8561 - 133rd Street, Surrey, British Columbia,
Canada V3W 4N8.
Ageless, Ageless Z, Ageless Z-200, Ageless-Eye, Mitsubishi Gas Chemical Co., Mitsubishi
Building, 5-2 Marunouchi 2-chome, Chiyoda-ku, Tokyo, 110 Japan. (Dierent types of Ageless are
available depending upon the water activity (WA) of the packaged commodity: AgelessZ WA
0.85%, Ageless A-200 indicates that 200 ml of oxygen can be absorbed. Ageless-Eye is used as a
color-changing oxygen indicator.)
Art-Sorb, Fuji Silysia Chemical Ltd., 6th Floor, YH Hisaya Building, 13-35, 1-Chome, Izumi,
Higashi-Ku, Nagoya-Shi, Aichi-Ken, 461 Japan.
Edney dial hygrometer, Edney 2 in dial hygrometer (ref. PH2P), M and T Precision Instruments
Ltd., Queensway, Eneld, Middlesex EN3 4SG, U.K.
Grant Squirrel Data Loggers, Grant Instruments Ltd., Barrington, Cambridge CB2 5QZ, U.K.
Humidical Corp. type card no. 6203-BB, Humidical Corp., 465 Mt. Vernon Avenue, P.O. Box 464,
Colton, CA 92324.
Kaken Gel, Kaken Pharmaceutical Co. Ltd., 2-28-8 Honkomagome, Bunkyo-ku, Tokyo, Japan.
Lexan, General Electric Plastics, Old Hall Road, Cheshire M33 2HG, U.K.
Materials and Suppliers
518 Wadum
Mirogard Protect Magic, Low-Iron, Deutsche Spezialglas AG, (DESAG/Schott), Postfach 2032,
31074 Grnenplan, Germany.
Nikka pellets, Nippon Kasseihakudo Co. Ltd. (Nippon Activated Clay Co. Ltd.), 7th Floor,
Daisan-Azuma Bldg., 1, Kandahirakawacho, Chiyoda-ku, Tokyo, 110 Japan.
Saran F-300, Dow Plastics, 2020 Willard Dow Center, Midland, MI 48674.
Squirrel, Eltek Ltd., 35 Barton Road, Haslingeld, Cambridge CB3 7LL, U.K.
Ashley-Smith, J., and A. J. Moncrie
1984 Experience with silica gel for controlling humidity in showcases. In ICOM Committee
for Conservation 7th Triennial Meeting, Copenhagen, 1014 September 1984, Preprints, vol. 2,
ed. Diana de Froment, 84.17.15. Paris: ICOM Committee for Conservation.
Bosshard, E.
1990 Klimavitrinen fr Gemlde, Eine wirksame und sthetisch befriedigende Methode.
Restauro 3:17680.
Bosshard, E., and M. Richard
1989 Climatized vitrines for paintings: An uncomplicated but ecient method. In American
Institute for Conservation: Annual Meeting Preprints, Cincinnati. Washington, D.C.: AIC.
Brewer, J. A.
1991 Eect of selected coatings on moisture sorption of selected wood test panels with
regard to common panel painting supports. Studies in Conservation 36:923.
Brimblecombe, P., and B. Ramer
1983 Museum display cases and the exchange of water vapors. Studies in
Conservation 28:17988.
Brown, J. P.
1994 Hygrometric measurement in museums: Calibration, accuracy, and the specication of
relative humidity. In Preprints of the Contributions to the Ottawa Congress, 1216 September
1994: Preventive ConservationPractice, Theory and Research, ed. Ashok Roy and Perry
Smith, 3943. London: IIC.
Buck, R. D.
1952 A note on the eect of age on the hygroscopic behavior of wood. Studies in
Conservation 1:3944.
1961 The use of moisture barriers on panel paintings. Studies in Conservation 6:919.
1962 Is cradling the answer? Studies in Conservation 7:7174.
1972 Some applications of rheology to the treatment of panel paintings. Studies in
Conservation 17:111.
1979 The Behavior of Wood and the Treatment of Panel Paintings: The 7th International Congress
of the International Institute for Conservation of Historic and Artistic Works, Wood in
Painting and the Decorative Arts, Oxford Congress 1978. Minneapolis: Upper Midwest
Conservation Association.
Byrne, R. O.
1984 An Easter Island egy gure display case. In ICOM Committee for Conservation 7th
Triennial Meeting, Copenhagen, 1014 September 1984, Preprints, vol. 2, ed. Diana de
Froment, 84.17.910. Paris: ICOM Committee for Conservation.
References
519 Mi crocii xart Boxts ror Pati Pai ri os
Cassar, M.
1984 Proposal for a typology of display case construction designs and museum climate
control systems. In ICOM Committee for Conservation 7th Triennial Meeting, Copenhagen,
1014 September 1984, Preprints, vol. 2, ed. Diana de Froment, 84.17.1115. Paris: ICOM
Committee for Conservation.
1985 Checklist for the establishment of a microclimate. The Conservator 9:1416.
1988 A microclimate within a frame for a portrait hung in a public place. In United Kingdom
Institute for Conservation 30th Anniversary Conference Preprints, comp. Victoria Todd,
4649. London: United Kingdom Institute for Conservation.
Cassar, M., and G. Martin
1994 The environmental performance of museum display cases. In Preprints of the
Contributions to the Ottawa Congress, 1216 September 1994: Preventive Conservation
Practice, Theory and Research, ed. Ashok Roy and Perry Smith, 17173. London: IIC.
Constable, W. G.
1934 Framing and the control of humidity. Appendix 3 of Some Notes on Atmospheric
Humidity in Relation to Works of Art, 48. London: Courtauld Institute of Art.
Curister, S.
1936 Control of air in cases and frames. Technical Studies in the Field of the Fine Arts 6:10916.
Daniels, V. D., and S. E. Wilthew
1983 An investigation into the use of cobalt salt impregnated papers for the measurement of
relative humidity. Studies in Conservation 28:8084.
de Guichen, G., and C. Kabaoglu
1985 How to make a rotten show-case. Museum 37(2):6467.
Diamond, M.
1974 A micro-micro-climate. Museums Journal 4:16163.
Edmunds, S.
1988 A microclimate box for a panel painting tted within the frame. In United Kingdom
Institute for Conservation 30th Anniversary Conference Preprints, comp. Victoria Todd,
5053. London: United Kingdom Institute for Conservation.
Erhard, D., and M. Mecklenburg
1994 Relative humidity re-examined. In Preprints of the Contributions to the Ottawa Congress,
1216 September 1994: Preventive ConservationPractice, Theory and Research, ed. Ashok
Roy and Perry Smith, 3238. London: IIC.
Feller, R. L.
1969 Transportation of a panel painting by courier in winter. In Papers Given at the Annual
Meeting of IIC-American Group, 1314. Los Angeles: IIC, American Group.
Gilberg, M.
1990 Inert atmosphere disinfestation using Ageless oxygen scavenger. In ICOM Committee for
Conservation 9th Triennial Meeting, Dresden, German Democratic Republic, 2631 August
1990, Preprints, vol. 2, ed. Kirsten Grimstad, 81216. Los Angeles: ICOM Committee
for Conservation.
Gilberg, M., and D. W. Grattan
1994 Oxygen-free storage using Ageless oxygen absorber. In Preprints of the Contributions to
the Ottawa Congress, 1216 September 1994: Preventive ConservationPractice, Theory and
Research, ed. Ashok Roy and Perry Smith, 17780. London: IIC.
520 Wadum
Hackney, S.
1987 The dimensional stability of paintings in transit. In ICOM Committee for Conservation 8th
Triennial Meeting, Sydney, Australia, 611 September 1987, Preprints, 597600. Marina del
Rey, Calif.: Getty Conservation Institute.
1990 Framing for conservation at the Tate Gallery. The Conservator 14:4452.
Kamba, N.
1993 Measurement of the dimensional change of wood in a closed case. In ICOM Committee
for Conservaton 10th Triennial Meeting, Washington, D.C., U.S.A., 2227 August 1993,
Preprints, vol. 1, ed. Janet Bridgland, 4069. Paris: ICOM Committee for Conservation.
Kenjo, T., and K. Toishi
1975 Purication of air with zeolite. Science for Conservation 12:2731.
Klein, P., and F. W. Brker
1990 Investigation on swelling and shrinkage of panels with wooden supports. In ICOM
Committee for Conservation 9th Triennial Meeting, Dresden, German Democratic Republic,
2631 August 1990, Preprints, vol. 1, ed. Kirsten Grimstad, 4145. Los Angeles: ICOM
Committee for Conservation.
Knight, P.
1983 An enclosure for the Tate panels in the Church of All Hallows Berkyngechirche by the
Tower. Conservator 7:3436.
Lafontaine, R. H., and S. Michalski
1984 The control of relative humidity: Recent developments. In ICOM Committee for
Conservation 7th Triennial Meeting, Copenhagen, 1014 September 1984, Preprints, vol. 2, ed.
Diana de Froment, 84.17.3337. Paris: ICOM Committee for Conservation.
Lambert, F. L., V. Daniel, and F. D. Preusser
1992 The rate of absorption of oxygen by Ageless: The utility of an oxygen scavenger in
sealed cases. Studies in Conservation 37:26774.
MacIntyre, J.
1934 Air conditioning for Mantegnas cartoons at Hampton Court Palace. Technical Studies in
the Field of Fine Arts 2(4):17184.
Mayer, M.
1988 Erfassung des Vitrinenklimas. In Mitteilungen des sterreichischen Restauratorenverbandes,
5156. Vienna: sterreichischer Restauratorenverband.
Mecklenburg, M. F., ed.
1991 Art in Transit: Studies in the Transport of Paintings. Washington, D.C.: National
Gallery of Art.
Mecklenburg, M. F., and C. S. Tumosa
1991 Mechanical behavior of paintings subjected to changes in temperature and relative
humidity. In Art in Transit: Studies in the Transport of Paintings, ed. M. F. Mecklenburg,
173216. Washington D.C.: National Gallery of Art.
Michalski, S.
1985 A relative humidity control module. Museum 37(2):8588.
1991 Paintings: Their response to temperature, relative humidity, shock, and vibration. In
Art in Transit: Studies in the Transport of Paintings, ed. M. F. Mecklenburg, 22348.
Washington D.C.: National Gallery of Art.
521 Mi crocii xart Boxts ror Pati Pai ri os
Miura, S.
1981 Studies on the behaviour of RH within an exhibition case. Part II: The static and
dynamic characteristics of sorbents to control the RH of a show-case. In ICOM
Committee for Conservation 6th Triennial Meeting, Ottawa, 2125 September 1981, Preprints,
81.18.5.110. Paris: ICOM Committee for Conservation.
Padeld, T.
1966 The control of relative humidity and air pollution in show-cases and picture frames.
Studies in Conservation 1:830.
Padeld, T., T. Burke, and D. Erhard
1984 A cooled display case for George Washingtons Commission. In ICOM Committee for
Conservation 7th Triennial Meeting, Copenhagen, 1014 September 1984, Preprints, vol. 2,
ed. Diana de Froment, 84.17.3842. Paris: ICOM Committee for Conservation.
Padeld, T., D. Erhard, and W. Hopwood
1982 Trouble in store. In Science and Technology in the Service of Conservation: Preprints to the
Washington IIC Congress, 39 September 1982, ed. N. S. Bromelle and G. Thomson,
2427. London: IIC.
Phillimore, E. A., ed.
1979 In Search of the Black Box: A Report on Proceedings of a Workshop on Micro-Climates Held at
the Royal Ontario Museum, February 1978. Toronto: Royal Ontario Museum.
Ramer, B. L.
1981 Stabilizing relative humidity variation within display cases: The role of silica gel and
case design. In ICOM Committee for Conservation 6th Triennial Meeting, Ottawa, 2125
September 1981, Preprints, 81.18.6.112. Paris: ICOM Committee for Conservation.
1984 The design and construction of two humidity-controlled display cases. In ICOM
Committee for Conservation 7th Triennial Meeting, Copenhagen, 1014 September 1984,
Preprints, vol. 2, ed. Diana de Froment, 84.17.4649. Paris: ICOM Committee for
Conservation.
1985 Show-cases modied for climate control. Museum 37(2):9194.
Ranacher, M.
1988 Zum Bau von Klimavitrinen mit direkter elektronischer Messung. In Mitteilungen
des sterreichischen Restauratorenverbandes, 4149. Vienna: sterreichischer
Restauratorenverband.
1994 The cold wall problem as a cause of accelerated aging of paintings. Poster presented
at International Institute for the Conservation of Historic and Artistic Works (IIC)
Ottawa Congress, 1216 September 1994.
Richard, M.
1991 Control of temperature and relative humidity in packing cases. In Art in Transit: Studies
in the Transport of Paintings, ed. M. F. Mecklenburg, 27997. Washington, D.C.:
National Gallery of Art.
1993 Conversation with the author. Conservation Laboratory, National Gallery of Art,
Washington, D.C.
1994 The transport of paintings in microclimate display cases. In Preprints of the
Contributions to the Ottawa Congress, 1216 September 1994: Preventive Conservation
Practice, Theory and Research, ed. Ashok Roy and Perry Smith, 18589. London: IIC.
Rothe, A., and B. Metro
1985 Climate controlled show-cases for paintings. Museum 37(2):8991.
522 Wadum
Sack, S. P.
196364 A case study of humidity control. Brooklyn Museum Annual 5:99103.
Sack, S. P., and N. Stolow
1978 A micro-climate for a Fayum painting. Studies in Conservation 23:4756.
Saunders, D., and T. Reeve
1993 Protective glass for paintings. National Gallery Technical Bulletin 15:98103.
Saunders, D., C. L. Sitwell, and S. Staniforth
1991 Soft pack: The soft option. In Art in Transit: Studies in the Transport of Paintings,
ed. M. F. Mecklenburg, 31121. Washington, D.C.: National Gallery of Art.
Schweizer, F.
1984 Stabilization of RH in exhibition cases: An experimental approach. In ICOM Committee
for Conservation 7th Triennial Meeting, Copenhagen, 1014 September 1984, Preprints, vol. 2,
ed. Diana de Froment, 84.17.5053. Paris: ICOM Committee for Conservation.
Schweizer, F., and A. Rinuy
1980 Zur Microklimatisierung zweier Vitrinen mit Ikonen fr eine temporre Ausstellung.
Maltechnik-Restauro 4:23943.
Shashoua, Y., and S. Thomson
1991 A eld trial for the use of Ageless in the preservation of rubber in museum collections.
In Canadian Conservation Institute Symposium 91, Saving the Twentieth Century: The
Degradation and Conservation of Modern Materials, Abstracts, 14. Ottawa: Canadian
Conservation Institute.
Simpson, W. S.
1893 An Improved Method of Means of Preserving Oil Paintings, Water Colour Drawings,
Engravings, Photographs, Prints, and Printed Matter from Atmospherical Deterioration and
from Decay. London: Her Majestys Stationery Oce.
Sozzani, L.
1992 Climate vitrine for paintings using the pictures frame as primary housing. Report,
Conservation Department, Rijksmuseum, Amsterdam.
Stamm, A. J.
1964 Wood and Cellulose Science. New York: Roland Press.
Stevens, W. C.
1961 Rates of change in the dimensions and moisture contents of wooden panels resulting
from changes in the ambient air conditions. Studies in Conservation 1:2124.
Stolow, N.
1965 Report on controlled environment for works of art in transit. Joint meeting of the
ICOM committee for Scientic Museum Laboratories and the ICOM sub-committee
for the Care of Paintings, Washington and New York. Typescript.
1966 Fundamental case design for humidity sensitive museum collections. Museum News
(Washington, D.C.) 44(6):4552.
1967 Standards for the care of works of art in transit. Report, International Institute for
the Conservation of Historic and Artistic Works London Conference on Museum
Climatology. Typescript.
1977 The microclimate: A localized solution. Museum News (Washington, D.C.) 56:5263.
1981 Controlled climate cases for works of art: Installation and monitoring. In ICOM
Committee for Conservation 6th Triennial Meeting, Ottawa, 2125 September 1981, Preprints,
81.18.8.14. Paris: ICOM Committee for Conservation.
523 Mi crocii xart Boxts ror Pati Pai ri os
Thomson, G.
1961 Museum climate: Humidity control, packing and transport. Studies in
Conservation 6:11011.
1964 Relative humidity: Variation with temperature in a case containing wood. Studies in
Conservation 9:15369.
1977 Stabilization of RH in exhibition cases: Hygrometric half-time. Studies in
Conservation 22:85102.
1981 Control of the environment for good or ill? Monitoring. National Gallery Technical
Bulletin 5:313.
Toishi, K.
1961 Relative humidity in a closed package. Studies in Conservation 6:11112.
Toishi, K., and S. Miura
1977 Purication of air with zeolite. Science for Conservation 14:17.
Toshiko, K.
1980 Studies on long-term conservation of cultural properties: The eect of dierent
concentration of oxygen on pigments used for cultural properties. Scientic Papers on
Japanese Antiques and Art Crafts 15:1037.
Wadum, J.
1992 De betere klimaatdoos. Het Mauritshuis Nieuwsbrief 2:89.
1993 Microclimate-boxes revisited. Talk given at the ICOM-CC meeting of the Working
Group on the Care of Works of Art in Transit, August, 1993, Washington, D.C.
Wadum, J., I. J. Hummelen, W. Kragt, B. A. H. G. Jtte, and L. Sozzani
1994 Research programme microclimates: Paintings on panel and canvas. Poster presented
at International Institute for the Conservation of Historic and Artistic Works (IIC)
Ottawa Congress, 1216 September 1994.
Weintraub, S.
1981 Studies on the behavior of RH within an exhibition case. Part I: Measuring the
eectiveness of sorbents for use in an enclosed showcase. In ICOM Committee for
Conservation 6th Triennial Meeting, Ottawa, 2125 September 1981, Preprints, 81.18.4.111.
Paris: ICOM Committee for Conservation.
1982 A new silica gel and recommendations. In American Institute for Conservation of Historic
and Artistic Works: Preprints of Papers Presented at the 10th Annual Meeting, Milwaukee,
Wisconsin, 2630 May 1982, 16973. Washington, D.C.: AIC.
Wilson, B. H., and L. W. Barridge
1933 Improvement in Controlling the Humidity of Air in Enclosed Spaces Such as Containers, Picture
Frames, and Rooms. London: Her Majestys Stationery Oce.
524 Wadum
vtr rnt tars, panel paintings have suered damage from a
wide range of causesaccidents, natural catastrophes, improper
handling, dramatic environmental changes, and misguided con-
servation treatments. Once damaged, panel paintings can be dicult to
repair. Due to this risk, many museum professionals and collectors are
hesitant to transport panels unless absolutely necessary. Some institutions
have even adopted policies that forbid their loan. In the United States,
panel paintings are not indemnied by the Arts and Artifacts Indemnity
program, a government program that provides insurance for international
exhibitions designated as being in the national interest.
Indeed, some paintings on wood supports are very fragile and
should not be transported or loaned to other institutions. Even the most
ideal packing case cannot protect a painting in very poor condition.
Many panel paintings are very stable, however, and can be safely packed
and transported.
A thorough technical examination of panel paintings considered
for loan is probably the most crucial aspect of the loan process. This exami-
nation is especially useful if condition and treatment records have been
maintained for many years. Paintings that have recurring problems such as
aking paint are poor candidates for loans, unless the cause of the insecu-
rity of the paint is clearly understood and controllable.
There are four environmental conditions that should be consid-
ered when evaluating any painting for possible loan: relative humidity
(RH), temperature, shock, and vibration. The overall safety of a painting
during transit is gauged by any expected response to these conditions; this
response must then be evaluated in terms of what the painting will be able
to withstand and what protection the proposed transport is able to pro-
vide. For example, a very fragile painting might suer impact poorly, and
no packing condition would be able to provide the protection needed to
ensure safe transport. If this is the particular case, transport of the paint-
ing is not recommended. However, if the painting can sustain moderate
uctuations in RH and temperature (factors easily controlled during trans-
port), and the panel can safely resist the anticipated levels of shock and
vibration, then the panel is a more likely candidate for loan.
There are several things to consider about the painting itself
when contemplating a possible loan, including the following: the size of
the painting, its materials and construction, the condition of the design
(paint and ground) layers, and the condition of the wood supports. Small
525
Mervin Richard, Marion Mecklenburg, and Charles S. Tumosa
Technical Considerations for the Transport of
Panel Paintings
O
paintings usually present fewer diculties than large paintings, since they
are lightweight, easily moved, and frequently made of a single piece of
wood. Large panels are heavier and more subject to bending moments
during handling operations, because of their own weight and width.
Bending or exing can also result from impact and vibration, which will
increase the stress throughout the panel and have particularly adverse
eects on poorly glued joints and existing cracks in the wood.
Considerable anecdotal evidence shows that some panels have been
exposed to extensive environmental uctuations for years without apparent
damage, while others subjected to similar conditions have suered. Some
paintings have remained stable for centuries, probably only because their
environment has also remained relatively stable. If subjected to a dierent
environment, the same paintings might rapidly develop problems.
Until recently, the only way to verify and observe this eect was
to change the environment to see what occurs. Obviously, this test can
prove destructive: damage has been reported when paintings have been
moved from relatively damp churches to drier and better-controlled envi-
ronments in museums or private homes. Similar problems also have devel-
oped when central heating systems without humidication have been
installed in buildings that were normally cold and damp. These reports
have led institutions to become cautious when considering the advisability
of lending a panel painting. Lenders to exhibitions frequently require that
borrowers maintain environmental RH levels closely matching the condi-
tions where their paintings are exhibited.
Battens or cradles have often been added to the reverse of panels,
either to reinforce the panels or to reduce warping. Usually such restora-
tion treatments have limited success and often lead to additional problems,
since these devices tend to restrain RH- and temperature-related move-
ment in the cross-grain direction of the panel. This restraint can lead to
excessive stresses (either compressive or tensile) if the RH or temperature
signicantly deviates from the conditions present when the battens or
cradle were applied.
The issue, then, lies in assessing the eects of changes in tempera-
ture and RH, as well as the events of impact and vibration on panel paint-
ings, and recognizing the limitations of controlling these factors during
transport. The typically short duration of transport usually precludes
chemical damage to paintings, but occasionally biological problems, such
as mold growth, arise. For the most part, determining the risks inherent to
the transport of a panel painting is an engineering problem that requires a
knowledge of the mechanics of artists materials. This particular discipline
is an important part of the authors current research, and a summary of
materials behavior is a signicant focus of this article.
All the materials typically found in panel paintings are hygroscopic; they
adsorb water when the RH increases and desorb water when the RH
decreases. These materials include the wood supports, hide glues, gesso
and paint layers, and varnishes. When these materials are unrestrained,
changes in their moisture content result in expansion and contraction. It
should be noted that panel materials respond dierently to the gain and
loss of water vapor. Oil paints and gessoes show relatively little dimen-
sional response to moisture, for example, as compared to pure hide glue or
to wood cut in the tangential direction. Wood cut in the radial direction
RH and Moisture Content
526 Ri chard, Me ckl e nburg, and Tumos a
shows about one-half of the dimensional response of wood cut in the tan-
gential direction (U.S. Department of Agriculture 1987). The dimensional
response of wood in the parallel-to-grain direction is 0.050.08% of that in
the tangential direction. In the tangential direction, some woods (e.g., cot-
tonwood [Populus spp.] and white oak [Quercus spp.]) can swell as much as
7% when subjected to changes from 5% to 95% RH. Other woods (e.g.,
spruce [Picea spp.] and mahogany [Swietenia macrophylla sp.]) swell only 3.5%
under similar conditions. The rate of dimensional change with respect to
RH is usually called the moisture coecient of expansion and is cited in units
of strain per percentage RH (mm/mm/% RH). It is of critical importance
to recognize that free-swelling dimensional changes are stress-free strains.
It is only when under restraint that hygroscopic materials subjected to RH
changes develop stress-associated strains. These are called mechanical
strains, in the truest sense of the word.
A coecient of expansion is often considered to be a constant;
however, the moisture coecients for these materials are not only variable
but highly nonlinear as well. In Figure 1, the moisture coecients for four
materials are plotted versus RH. These materials are a fteen-year-old
ake white oil paint, gesso with a pigment volume concentration of
81.6%, hide glue, and a sample of white oak in the tangential direction.
In this plot, the longitudinal direction of the white oak (or of any wood)
would factor almost along the zero line. In Figure 1 all of the materials
have very low rates of dimensional response with respect to RH in the
4060% range. Outside this range the wood and glue show dramatic
increases in the rate of dimensional response with respect to RH, and
there is a signicant deviation of the wood and glue responses in relation
to the paint and gesso responses. This mismatch in the coecients is
indicative of the source of most of the problems associated with environ-
mental changes. Wood in the longitudinal direction responds much less to
the environment than do the paint and gesso, which essentially means that
dierent responses are occurring to the paintings layers in the two perpen-
dicular directions of the panel. The responses of the materials to RH can
be studied either alone or as part of a composite construction.
A material that is allowed to expand and contract freely can be
repeatedly subjected to a fairly wide RH range without damage. In addi-
tion, woods (e.g., white oak) show a dramatic hysteresis when the unre-
strained dimensional response is measured over a very large range of
humidity. The increasing RH path tends to stay lower than the decreasing
RH path; therefore, if the measurements are taken at 2575% RH, the
increasing and decreasing paths are almost the same.
A structural problem arises when either full or partial restraint is
present. This restraint can result from defects such as knots in the wood,
cross-grain construction (often found in furniture), or battens that are
attached to the reverse of a panel. If battens and cradles restrict the
dimensional movement of the wood, stresses and strains develop perpen-
dicular to the grain with changes in RH. Internal restraint can develop
when the outer layers of a massive material respond more quickly than
the interior layer.
Research has shown that there are reversible levels of stress and
strain. In the case of a fully restrained material (white oak in the tangential
direction, for example), some changes in RH can occur without ill eect to
the wood (Mecklenburg, Tumosa, and Erhardt 1998). Organic materials
(i.e., wood, paints, glue, gesso) have yield points, which are levels of strain
527 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
below full reversibility and above permanent deformation. Measured by an
axial mechanical test, the initial yield points for woods, paints, and glues are
approximately 0.004. These materials can, however, harden under strain, a
process that creates substantial increases in their yield points. For a brittle
gesso found in a traditional panel painting, the yield point is approximately
0.0025. If gessoes are richer in glue, both their yield points and their strains
at failure increase signicantly. The magnitudes of yield points do not
appear to be appreciably aected by RH, but generally the strains to break-
ing will increase parallel to increases in RH. Finally, RH- and temperature-
related events are biaxial and triaxial events. This means that yielding can
occur at signicantly higher strain levels than axial testing would indicate.
In this article, the lowest axially measured strain level of 0.004 will be used
for all materials except gesso, which yields at 0.0025. These yield points will
be used to determine the maximum allowable RH uctuations in panels.
This approach is a fairly conservative one to assessing the eects of RH and
temperature on panel paintings, and it should be considered accordingly. It
also should be noted here that while materials yield at strains of 0.004 or
greater between 35% and 65% RH, strains of 0.009 or greater are necessary
to cause failure. The strains at failure in seriously degraded materials are
often lower because the process of degradation usually reduces strength.
When the magnitude of the failure strains approaches that of the yield
strains, the materials of the panel painting are considered fragile and proba-
bly dicult to handle, as they will break in an elastic region rather than
plastically deform.
Response of restrained wood to RH: Tangential direction
Research has shown that the moisture coecient of a material can be used
to calculate the RH change required to induce both yielding and failure
strains in a restrained material (Mecklenburg, Tumosa, and McCormick-
Goodhart 1995). Equation 1 shows how these mechanical strains can be
calculated as a function of RH. Using this equation, the strain change ()
for any RH change can be calculated by integrating from one RH point to
another as
dRH (1)
where: d/dRH, the moisture coecient of expansion.
The yield point for white oak is about 0.004 at all RH levels, and
its breaking strains increase with increasing RH. These strain values are
shown in Figure 2. The failure strains are small at a low RH and increase
dramatically as RH increases.
With the information from Figures 1 and 2 and Equation 1, it is
possible to develop a picture of the eects of RH on the strains of white
oak fully restrained in the tangential direction. This is a hypothetical
example of the worst condition possible; fortunately, few objects in collec-
tions are actually fully restrained. The plotted results of calculations made
using Equation 1 are shown in Figure 3. In this plot, the calculated results
show what would occur if white oak in the tangential direction were
restrained at 50% RH, then subjected to RH changes. A decrease to
approximately 33% RH would result in tensile yielding of the wood.
Further decreasing, to 21% RH, could cause the wood to crack. Increasing
the RH from 50% to approximately 64% would cause the wood to begin
528 Ri chard, Me ckl e nburg, and Tumos a
compression yielding. As long as the RH remains between approximately
33% and 64%, the wood can respond dimensionally without its structure
being altered. However, if the RH increases above approximately 64%,
compression set may occur, which is a permanent deformation of the wood.
Compression set also re-initializes the wood to a new, higher RH environ-
ment, causing the wood to behave like one acclimated to a higher RH.
The plots in Figure 4 were obtained by recalculating Equation 1 for
the fully restrained white oak panel, now acclimated to 70% RH (the cir-
cumstances under which the panel acclimated to a higher ambient RH are
irrelevantit does not matter whether the painting has always been main-
tained at 70% or whether it was temporarily stored in a damp location).
529 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
0.0050
0.0045
0.0040
0.0035
0.0030
0.0025
0.0020
0.0015
0.0010
0.0005
0
0 20 40 60 80 100
Gesso
Flake white
oil paint
RH (%)
M
o
i
s
t
u
r
e
c
o
e
f
f
i
c
i
e
n
t
o
f
e
x
p
a
n
s
i
o
nWhite oak
tangential direction
Hide glue
0.040
0.035
0.300
0.025
0.020
0.015
0.010
0.005
0
0 20 40 60 80 100
RH (%)
Yield
Failure
S
t
r
a
i
n
Fi gure 1
Moisture coecients of expansion versus RH
for four materials: white oak in the tangential
direction, hide glue, gesso, and fteen-year-
old ake white oil paint. The radial-direction
coecient for white oak is approximately one-
half of the tangential, and the longitudinal-
direction coecient is about one-tenth of the
tangential. The swelling rate is the lowest in
the midrange RH levels.
Fi gure 2
Measured yield and breaking strains of
tangential-direction white oak versus RH
(axial tensile test).
A problem becomes apparent when desiccation of the panel is attempted.
A drop from 70% to 62% RH causes tensile yielding, and a drop to approxi-
mately 38% RH can cause cracking of the wood. Increasing the RH to
approximately 74% induces yielding in compression. The panel cannot
tolerate the much larger variations in RH that are possible with a panel
equilibrated to 50% RH, as seen in Figure 3. This narrow range of RH
must be considered when evaluating the risks of lending panel paintings
acclimated to high RH.
In the past, some panels have been treated with water or large
amounts of water vapor in an attempt to atten them. Battens or cradles
530 Ri chard, Me ckl e nburg, and Tumos a
Starting RH (%)
0 20 40 60 80 100
110
100
90
80
70
60
50
40
30
20
10
0
E
n
d
i
n
g
R
H
(
%
)
Restrained at this RH
Tension response
Compression response
Yield 0.004
Yields in compression at this RH
(compression set begins)
Zero stress or strain
Yields in tension at this RH
Cracks in tension at this RH
Yield 0.004
Failure
110
100
90
80
70
60
50
40
30
20
10
0
0 20 40 60 80 100
Compression response
Starting RH (%)
E
n
d
i
n
g
R
H
(
%
)
Yield 0.004
Yield 0.004
Failure
Tension response
Restrained at this RH
Cracks in tension at this RH
Yields in tension at this RH
Zero stress or strain
Yields in compression at this RH
(compression set begins)
Fi gure 3
Calculated reversible RH range of fully
restrained, tangentially cut white oak versus
ambient RH. A yield value of 0.004 was used
as the limiting criterion in both tension and
compression. The values of the dotted lines
are for stress-free wood that has been fully
equilibrated to 50% RH.
Fi gure 4
Calculated reversible RH range of fully
restrained, tangentially cut white oak versus
ambient RH. A yield value of 0.004 was used
as the limiting criterion in both tension and
compression. The wood has been fully equili-
brated to 70% RH. The allowable RH range
has been severely reduced in comparison to
wood equilibrated to 50% RH.
were often attached to the reverse while a panel was still wet. The eect of
this treatment was to restrain the panel while it was still acclimated at an
extremely high RH. As the panel dries, the adhesive hardens, and the point
of full restraint could easily have a moisture content equivalent to acclima-
tion of the wood at 75% RH. If this is the case, this panel will yield in ten-
sion at around 68% RH and could quite possibly crack at approximately
45% RH. If a restrained panel were to be subjected to a ood (such as
occurred in Florence in 1966), the simple act of drying would be almost
certain to cause wood-support damage unless all of the restraint were
removed before drying.
Figure 5 shows the results of RH uctuations on a typical white
oak panel restrained and equilibrated at 36% RH. In this case the panel
will yield in compression at approximately 53% RH and in tension at 25%
RH. The eect is to simply ensure that the reversible environment for the
painting support panels is changed to a lower RH.
For comparison purposes, the moisture coecient of expansion
for a 100-year-old white oak sample was measured in the tangential direc-
tion. This measurement allows for a comparison of the strain development
in new and aged oak. Figure 6 shows that when the same yield criterion
(0.004) is used, the 100-year-old oak appears to be able to sustain slightly
greater RH variations, particularly at the extreme ranges of the RH spec-
trum. Many other woods used as painting supports have less dimensional
response to moisture than white oak, so their allowable uctuations will
be signicantly greater, even in the tangential grain direction.
Response of restrained wood to RH: Radial direction
The moisture coecient of expansion in the radial direction is about one-
half that of the tangential direction. If a wood panel support is made so
that the two primary directions of the wood are longitudinal and radial,
the panel can sustain signicantly greater variations in humidity than if a
primary direction were tangential. Figure 7 shows a comparison of the
531 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
110
100
90
80
70
60
50
40
30
20
10
0
0 20 40 60 80 100
Compression response
Starting RH (%)
E
n
d
i
n
g
R
H
(
%
)
Yield 0.004
Yield 0.004
Failure
Tension response
Restrained at this RH
Cracks in tension at this RH
Yields in tension at this RH
Zero stress or strain
Yields in compression at this RH
(compression set begins)
Fi gure 5
Calculated reversible RH range of fully
restrained, tangentially cut white oak versus
ambient RH. A yield value of 0.004 was used
as the limiting criterion in both tension and
compression. The wood has been fully equili-
brated to 36% RH. The allowable RH range
is still fairly broad, but it has been shifted
to lower values.
calculated RH changes required to reach yield in both the radial and tangen-
tial directions for 100-year-old white oak. If it is assumed that the panels
had been restrained at 50% RH, the RH change required to cause yielding
in tension is a decrease to 31% in the tangential direction and to 23% in
the radial direction. An increase in RH to 65% would cause compressive
yielding in the tangential direction; an increase in RH to 75% would cause
compressive yielding in the radial direction. Because of its substantial
increase in the allowable changes in RH, radial cutting is an important
consideration for woods that are to be acclimated and restrained at high
RH. In Figure 8 the restrained panels are shown as equilibrated to 70% RH.
In the radial direction the wood would be capable of sustaining a drop to
532 Ri chard, Me ckl e nburg, and Tumos a
110
100
90
80
70
60
50
40
30
20
10
0
0 20 40 60 80 100
Compression response
Starting RH (%)
E
n
d
i
n
g
R
H
(
%
)
New-oak yield 0.004
New-oak yield 0.004
New-oak tensile failure
Tension response
Restrained at this RH
100-year-old oak yield 0.004
100-year-old oak yield 0.004
Fi gure 6
Calculated reversible RH range of fully
restrained, new, tangentially cut white oak
versus ambient RH, compared to 100-year-old
oak. A yield value of 0.004 was used as the
limiting criterion in both tension and com-
pression. It is assumed that the wood has been
fully equilibrated to 50% RH.
110
100
90
80
70
60
50
40
30
20
10
0
0 20 40 60 80 100
Starting RH (%)
E
n
d
i
n
g
R
H
(
%
)
Radial yielding
Tangential yielding
Tension response
Restrained at this RH
Compression response
All yield strains 0.004
Tangential yielding
Radial
yielding
Fi gure 7
Calculated reversible RH range of fully
restrained, 100-year-old, radially cut white oak
versus ambient RH, compared to 100-year-old
tangentially cut oak. A yield value of 0.004
was used as the limiting criterion in both ten-
sion and compression. It is assumed that the
wood has been fully equilibrated to 50% RH.
The signicant increase of allowable RH in the
radial direction demonstrates the advantages
of preparing panel supports in that direction.
533 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
40% RH before yielding in tension, and capable of sustaining an increase
to 86% RH before compression set begins. In the tangential direction, the
panel is restricted to a range of 5579% RH. The implications of these
results are clear: panels cut in the tangential direction present a signicantly
greater risk of movement, particularly if acclimated to a high RH. In con-
trast, restrained panels cut in the radial direction are low risk, even if they
have been acclimated to 70% RH.
The above examples help illustrate the response of wood to RH.
Knowledge of the history, wood type, treatment record, and grain orienta-
tion of a panel painting is highly useful in helping to determine its poten-
tial risk from changes in RH and its subsequent potential for safe travel.
This study used the extremes of conservative yield criteria and assump-
tions of worst-case full restraint.
Response of the design layers to RH
Until now, only the wooden panel has been discussed. However, it is also
important to examine other components of the panel, such as gesso and
oil paint layers. Since paint and gesso have very similar dimensional
responses to changes in RH over most of the RH range, similar eects
will occur when these layers are considered as coatings on panels that are
both restrained and unrestrained (i.e., without battens, cradles, or fram-
ing techniques).
The primary dierence between the two materials is that paint
will be assumed to yield at a strain of 0.004 and gesso at a strain of about
0.0025. Therefore, while gesso and paint do have similar dimensional
responses to changes in RH, the gesso will yield sooner to those changes
than will the paint. As was seen with the wood, once paint or gesso is
beyond the yield point, nonreversible strains occur. Depending on the
environment to which the panel is acclimated, damage can be anticipated
if the equilibrated RH deviations are well in excess of those causing yield-
ing. Since not all paintings have gesso layers, the following comments will
100
90
80
70
60
50
40
30
20
10
0
0 20 40 60 80 100
Starting RH (%)
E
n
d
i
n
g
R
H
(
%
)
Radial yielding
Tangential yielding
Tension response
Restrained at this RH
Compression response
All yield strains 0.004
Tangential yielding
Radial yielding
Fi gure 8
Calculated reversible RH range of fully
restrained, 100-year-old, radially cut white oak
versus ambient RH, compared to 100-year-old
tangentially cut oak. A yield value of 0.004
was used as the limiting criterion in both ten-
sion and compression. It is assumed that the
wood has been fully equilibrated to 70% RH.
The signicant increase of allowable RH in
the radial direction demonstrates the advan-
tages of preparing panel supports in that
direction. This consideration is particularly
important in the case of panels equilibrated
to high RH levels.
534 Ri chard, Me ckl e nburg, and Tumos a
distinguish between the eect of RH on panels having both gesso and paint
layers and the effect on panels having paint directly applied to the wood.
Unrestrained wooden panels in the tangential direction exhibit
substantial dimensional uctuations with RH changes. If the swelling
coecients of expansion of all materials applied to the wood panel are
the same as those of the wood, then RH variations will induce no stresses
in the attached layers. If the swelling coecients dier, mechanical stresses
and strains will develop as a result of RH changes. For example, in the
longitudinal direction of a panel painting, the wood is minimally respon-
sive to RH. The paint and gesso coatings are responsive, but the wood
restrains these layers from shrinking and swelling with changes in RH. In
the tangential direction, however, the wood is much more responsive to
RH variations than the gesso or paint. The responsiveness of the wood
also creates stresses and strains in the design layers. In eect, the wood is
overriding the response of the design layers.
The mechanical strains in the paint and gesso layers can be calcu-
lated using Equation 2. This equation can be used for any material applied
to any substrate, provided the substrate is substantially thicker than the
applied layers. (To check this equation, assume that the coecient of
expansion for the substrate is zero; Equation 2 would then simplify to
Equation 1.) Equation 2 is

p
[(1
s
dRH) (1
p
dRH)]/(1
s
dRH) (2)
where:
s
is the swelling coecient of the substrate, which is thick relative
to any attached layers; and
p
is the swelling coecient of the coatings,
either ake white paint or gesso. In our examples white oak is the substrate.
Response of the design layers to RH: Panels cut in the
tangential direction
In Figure 9 the calculated mechanical strains for ake white oil paint and
gesso (calcium carbonate and hide glue) on an unrestrained white oak
panel are plotted versus RH. The paint, gesso, and wooden support panel
are considered to be equilibrated to 50% RH, with initial stresses and
strains of zero. The strains are plotted versus RH in both the tangential
and longitudinal directions of the wooden panel support. In the longitudi-
nal direction, the wood acts as a full restraint to the applied coatings (paint
and gesso), and strains remain low over most of the RH range. The oil
paint and gesso are minimally responsive to moisturefor the paint, the
plot shows that it is possible to desiccate from 50% to 8% RH before ten-
sile yielding occurs. Compressive yielding in the paint occurs when the RH
is raised from 50% to approximately 95% (note that the paint is yielding,
not breaking). However, in the gesso (which yields at a lower strain), the
range for acceptable RH is narrower. In this case, tensile yielding will
occur at approximately 19% RH, and compressive yielding at approxi-
mately 83% RH. This indicates that fairly large RH variations can occur
without yield in the design layer. However, it is well known that cracks
do develop perpendicular to the grain of the wood, indicating that the
stresses and strains are parallel to the grain. This study shows that these
cracks do not usually occur as a result of moderate RH changes. Drops
in temperature are more likely to cause these types of cracks, as will be
discussed below.
As it responds to the moisture changes, the wooden substrate
signicantly aects the mechanical strains in both the paint and the gesso
layers. The strains of the design layers actually become compressive with
desiccation, because the wood shrinks at a greater rate than either the
paint or gessothe gesso yields at 33% RH, and the paint yields at 27%
RH. Further desiccation from the yield points causes permanent deforma-
tion in both layers. If the desiccation continues below 15% RH and the
gesso ground is not rmly attached, crushing may occur, and cleavage
ridges will develop parallel to the grain.
Raising the RH above 50% causes a dierent problem. At approxi-
mately 62% RH, the gesso begins to yield in tension; at about 65% RH,
the paint begins to yield in tension. At about 75% RH or above, strains in
the design layer can be high enough to induce cracking in a brittle gesso
layer. This cracking of the gesso can subsequently crack the paint lm
applied above it. These cracks appear parallel to the grain of the wooden
support panel. If no gesso layer is present, paint cracking would not begin
until well above 85% RH.
Diagrams similar to that in Figure 9 demonstrate the response of
gesso and paint layers attached to the panel when they are equilibrated to
RH levels other than 50%. Figure 10 shows the calculated resulting strains
developed in the paint and gesso when the panel painting has been equili-
brated to 64% RH. Tensile yielding in the paint now occurs at about 43%
RH (higher than when the painting was acclimated to 50% RH). At 53%
RH the gesso yields in tension. A 14% variation (5064% RH) in the equi-
librium environment will have a major eect on the dimensional response
of the panel. This panel is to some degree restricted to a narrower and
higher environment, as compared to a panel equilibrated to 50% RH. If,
however, the equilibrium environment is higher (e.g., about 70%), greater
dierences will occur in the response of the panel to the environment.
This is illustrated in Figure 11, which shows the calculated strains of the
design layers applied to a panel equilibrated to 70% RH. Under the condi-
535 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
0.016
0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
0.002
0.004
0.006
0.008
0.010
0.012
0.014
0.016
0 20 40 60 80 100
RH (%)
S
t
r
a
i
n
i
n
g
e
s
s
o
a
n
d
o
i
l
p
a
i
n
t
l
a
y
e
r
s
Longitudinal
direction
Equilibrium
environment
Tangential direction
0.004
0.004
0.0025
0.0025
Cleavage of
design layer
Cracking of
design layer
Fi gure 9
Calculated strains in gesso and ake white oil
paint applied to an unrestrained, tangentially
cut white oak panel versus RH. The panel
painting is assumed to be equilibrated to 50%
RH. Both the gesso and paint have fairly large
allowable RH uctuations, even in the tangen-
tial direction of the wood.
tions in this example, the gesso layer will yield with a drop in RH from
70% to 64%, and the paint will yield when the RH drops to 60%. Crushing
or cleavage of the design layer could occur at about 35% RH if the gesso
ground is not sound. A panel equilibrated to a high level of RH will suer
some permanent deformation if subjected to the well-controlled environ-
ments found in many institutions. In addition, a smaller increase in RH,
(68%), is needed to cause tensile yielding when compared to a panel
equilibrated to 50% RH.
How realistic is the example above? At such a high RH level, there
is a strong potential for biological attack that should be observed and
noted. For a panels RH to equilibrate to a high annual mean, RH levels
during the more humid periods of the year must also be high. Evidence of
536 Ri chard, Me ckl e nburg, and Tumos a
0 20 40 60 80 100
RH (%)
S
t
r
a
i
n
i
n
g
e
s
s
o
a
n
d
o
i
l
p
a
i
n
t
l
a
y
e
r
s
Longitudinal
direction
Equilibrium
environment
Tangential direction
0.004
0.004
0.0025
0.0025
Cleavage of
design layer
Cracking of
design layer
0.016
0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
0.002
0.004
0.006
0.008
0.010
0.012
0.014
0.016
0 20 40 60 80 100
RH (%)
S
t
r
a
i
n
i
n
g
e
s
s
o
a
n
d
o
i
l
p
a
i
n
t
l
a
y
e
r
s
Longitudinal
direction
Equilibrium
environment
Tangential direction
0.004
0.004
0.0025
0.0025
Cleavage of
design layer
Cracking of
design layer
0.016
0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
0.002
0.004
0.006
0.008
0.010
0.012
0.014
0.016
Fi gure 10
Calculated strains in gesso and ake white oil
paint applied to an unrestrained, tangentially
cut white oak panel versus RH. The panel
painting is assumed to be equilibrated to 64%
RH. The paint still has a fairly large allowable
RH uctuation, even on the tangentially cut
wood, but the gesso is now conned to a
more restricted RH range.
0 20 40 60 80 100
RH (%)
S
t
r
a
i
n
i
n
g
e
s
s
o
a
n
d
o
i
l
p
a
i
n
t
l
a
y
e
r
s
Longitudinal
direction
Equilibrium
environment
Tangential direction
0.004
0.004
0.0025
0.0025
Cleavage of
design layer
Cracking of
design layer
0.016
0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
0.002
0.004
0.006
0.008
0.010
0.012
0.014
0.016
Fi gure 11
Calculated strains in gesso and ake white oil
paint applied to an unrestrained, tangentially
cut white oak panel versus RH. The panel
painting is assumed to be equilibrated to
70% RH. Both the gesso and paint are now
conned to a very restricted RH range in the
tangential direction. This painting would be at
serious risk if subjected to low RH levels.
mold damage could be an important indication that a panel painting may
have equilibrated to an excessively high humidity and therefore is a less-
than-suitable candidate for shipment.
If a panel painting has equilibrated to an environment lower than
50%, the RH changes needed to cause yielding are not signicantly
aected. Figure 12 shows the calculated results for a painting equilibrated
to 36% (rather than 50%) RH. Note that with a 14% downward shift in the
equilibrium environment, only about a 6% downward shift in the RH is
necessary to attain compressive yielding in both the gesso and the paint
layers. The panel painting equilibrated to this low-RH environment can
still sustain signicant deviations in the mid-RH range without yielding. In
addition, the painting has to drop to 26% RH for yielding in the gesso to
occur, and to 22% RH for yielding in the paint to occur.
Response of the design layers to RH: Panels cut in the
radial direction
Paintings executed on radially cut wooden panels are at reduced risk dur-
ing transport, and the layers applied to such panels are far less likely to
suer RH-related damage. Figure 13 illustrates the dierent responses of
the design layer to the unrestrained movement of white oak. In the longi-
tudinal direction, there is little dierence between tangentially and radially
prepared panels, and the strains in the gesso and paint layers are similar to
those shown in Figure 9. (As before, the assumed yield strains are 0.004 for
the paint and 0.0025 for the gesso.)
In a panel cut in the radial direction and acclimated to 50% RH,
gesso shows compressive yielding at 22% and shows tensile yielding at
79%. In a panel cut in the tangential direction, the gesso shows compres-
sive yielding at 33% RH and tensile yielding at 63% RH. If there is no
gesso layer, the paint lm attains compressive yielding at 13% RH and ten-
sile yielding at 86% RH. These RH values are not substantially reduced
from the RH yield points of the paint in the longitudinal direction. The
537 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
0 20 40 60 80 100
RH (%)
S
t
r
a
i
n
i
n
g
e
s
s
o
a
n
d
o
i
l
p
a
i
n
t
l
a
y
e
r
s
Longitudinal
direction
Equilibrium
environment
Tangential direction
0.004
0.004
0.0025
0.0025
Cleavage of
design layer
Cracking of
design layer
0.016
0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
0.002
0.004
0.006
0.008
0.010
0.012
0.014
0.016
Fi gure 12
Calculated strains in gesso and ake white oil
paint applied to an unrestrained, tangentially
cut white oak panel versus RH. The panel
painting is assumed to be equilibrated to
36% RH. Both the gesso and paint have large
allowable uctuations of RH, even in the
tangential direction. This painting would not
be at risk unless it were subjected to RH levels
above 55%.
538 Ri chard, Me ckl e nburg, and Tumos a
dierence is that with desiccation, the paint and gesso experience compres-
sion in the cross-grain direction and tension in the longitudinal direction,
while increases in humidity induce the opposite reaction. Both the wood
and the design layers are more stable on a radially cut panel.
Of signicant interest is the response of the design layers that
have been applied to radially cut oak and equilibrated to a high RH. In
Figure 14, the calculated strains in the paint and gesso layers applied to
radially cut oak and equilibrated to 70% RH are given. When desiccation
occurs, compressive yielding occurs in the gesso at 32% RH and in the
paint at 19% RH. Upon equilibration to 50% RH, tensile yielding in
the gesso occurs at 85% RH and in the paint at 90% RH. This is a sub-
0 20 40 60 80 100
RH (%)
S
t
r
a
i
n
i
n
g
e
s
s
o
a
n
d
o
i
l
p
a
i
n
t
l
a
y
e
r
s
Longitudinal
direction
Equilibrium
environment
Tangential
direction
0.004
0.004
0.0025
0.0025
Radial
direction
0.016
0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
0.002
0.004
0.006
0.008
0.010
0.012
0.014
0.016
Fi gure 13
Calculated strains in gesso and ake white oil
paint applied to unrestrained, radially and tan-
gentially cut white oak panels versus RH. The
panel paintings are assumed to be equilibrated
to 50% RH. Both the gesso and paint have
large allowable uctuations of RH, even in
the tangential direction, but the radial direc-
tion shows a signicant increase in the allow-
able uctuations over the tangential cut.
0 20 40 60 80 100
RH (%)
S
t
r
a
i
n
i
n
g
e
s
s
o
a
n
d
o
i
l
p
a
i
n
t
l
a
y
e
r
s
Longitudinal
direction
Equilibrium
environment
Tangential
direction
0.004
0.004
0.0025
0.0025
Radial
direction
0.016
0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
0.002
0.004
0.006
0.008
0.010
0.012
0.014
0.016
Fi gure 14
Calculated strains in gesso and ake white
oil paint applied to unrestrained, radially and
tangentially cut white oak panels versus RH.
The panels are assumed to be equilibrated to
70% RH. Both the gesso and paint have very
small allowable uctuations of RH in the
tangential direction, but the radial direction
shows a signicant increase in the allowable
uctuations. Where the tangentially cut panel
is at risk when equilibrated to high RH, the
radially cut panel can still sustain large RH
uctuations.
stantial improvement over the strains that developed in the design layers
that were applied to tangentially cut wood. Panels cut tangentially and
equilibrated to a high RH are at serious risk if desiccated. Panels cut radi-
ally are at considerably less risk, even when desiccated and equilibrated
to a high RH. For example, paintings on plywood panels that are made
entirely of restrained, tangentially cut wood fare poorly when exposed to
RH uctuations, as compared to paintings on radially cut panels, whether
restrained or not.
The equilibrium RH of a panel paintings environment establishes
its risks for transport. Knowing the equilibrium RH allows for the develop-
ment of environmental guidelines for both the transit case and the new,
temporary exhibition space. Tangentially cut panels acclimated to high RH
are at risk. This risk can occur when warped panels have been attened
with moisture before the addition of battens or cradles. In such instances a
warped panel is often thinned, moistened on the reverse, and nally
attached to battens or a cradle to forcibly hold the panel at. As a result,
considerable tensile stress can build up as the wood dries, since the battens
or constricted cradles can restrict the return to warpage.
When panels are thinned, there are other consequences.
Decreasing the thickness reduces the bending stiness of a panel and
makes it more exible. The reduction in stiness is inversely proportional
to the cube of the thickness of the panel (Weaver and Gere 1965:11517).
This thinning makes the panel prone to buckling when restrained. At a
high RH, a panel with a locked-in cradle is subjected to high RH-induced
compressive stresses in the spans between the cradle supports, and because
of the cradle, such stresses are not uniform. They cause out-of-plane bend-
ing or buckling of thinned panels.
It is important to assess whether a panels movement is
restrictedan assessment that may be dicult in some cases. Panels with
battens or cradles that have locked up by friction present higher risks for
transport if they are cracked or if the panel has equilibrated to a very high
RH environment (Mecklenburg and Tumosa 1991:18788). In addition,
research suggests that an unrestrained panel with a gesso layer equilibrated
to a high-RH environment is at greater risk of damage upon desiccation
than is a sound (free of cracks), restrained panel. This risk occurs because
the gesso layer is subject to compression cleavage when an unrestrained
panel contracts from desiccation. Almost all the panel paintings of the
fteenth- and sixteenth-century Italian Renaissance have gesso grounds.
This gesso layer and the wood panel itself should be considered the crucial
components when the movement of such paintings is contemplated.
In contrast, oil paintings on copper supports seem to have fared
well over the centuries. Research shows that oil paint responds only mod-
erately to changes in RH, particularly if extremely high RH levels are
avoided. Additionally, copper is dimensionally unresponsive to RH
uctuations. The combination of these two materials results in a painting
that is durable with respect to changes in atmospheric moisture.
Contemporary panel paintings having wooden supports and either
acrylic or alkyd design layers may also be analyzed in relation to the crite-
ria discussed above. Figure 15 shows the coecients for swelling of alkyd
and acrylic emulsion paints compared to those of oil paint. All of these
paints have dried for fteen years or more under normal drying condi-
tions. Both the alkyd and the acrylic emulsion paints are much less dimen-
sionally responsive to moisture than is oil paint. When acrylic paints are
539 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
540 Ri chard, Me ckl e nburg, and Tumos a
applied to a wooden panel, RH changes have very little eect in the longi-
tudinal direction of the wood. In the tangential direction, the movement
of the paint is almost totally dictated by the movement of an unrestrained
wooden panel. However, the RH change needed to develop yield in alkyd
or acrylic paints will be approximately 23% less than the change needed
for oil paint on wooden panels because the moisture coecient of expan-
sion of the oil paint is higher.
Control of transport RH
RH levels may also vary during transport, but fortunately this problem
can be solved with proper packing. Since the RH levels in trucks depend
largely on weather conditions, the RH inside even an air-conditioned truck
may be very high on a hot, humid day. If the weather is very cold, the RH
in the truck may be low because of the drying eects of the cargo-area
heating system. At high altitudes, the RH in a heated and partially pressur-
ized aircraft cargo space is always lowoften as low as 1015%. Panel
paintings exposed to this extreme desiccation for the duration of an aver-
age ight could be damaged. This desiccation can be avoided if the paint-
ing is wrapped in a material that functions as a moisture barrier (wrapping
of panel paintings is discussed further below).
The dimensional response of wooden panels to temperature variations has
been largely ignored by many conservators, because temperature has been
considered to have a much smaller eect on wood than has RH. This pre-
cept holds true if one considers only the relative dimensional response of
wood to temperature as compared to its response to moisture. It would
take a change of several hundred degrees in temperature to induce the
same dimensional change in wood that can be caused by a large change in
RH. Panel paintings are rarely exposed to such temperature extremes, and
they are usually exhibited or stored where temperature variations are rela-
tively small. The problem, however, is not so much the response of the
Temperature Eects
0 20 40 60 80 100
RH (%)
M
o
i
s
t
u
r
e
c
o
e
f
f
i
c
i
e
n
t
o
f
e
x
p
a
n
s
i
o
n
Cadmium yellow
alkyd medium
Cadmium yellow
safflower oil
Cobalt blue
acrylic emulsion
0.0016
0.0015
0.0014
0.0013
0.0012
0.0011
0.0010
0.0009
0.0008
0.0007
0.0006
0.0005
0.0004
0.0003
0.0002
0.0001
0.0000
Fi gure 15
Moisture coecients of expansion versus RH
for oil, alkyd, and acrylic paints. The dimen-
sional responses of the alkyd and acrylic
paints are substantially lower than those of
the oil paint.
541 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
wood as it is the response of the gesso and paint layers. Therefore, when
the eects of temperature are considered, it is also necessary that the
mechanical properties of the dierent paint media, as well as their dimen-
sional responses, are understood. In the temperature ranges most likely to
be encountered, the thermal coecients of expansion for the materials
found in panel paintings can easily be considered as constants. Some val-
ues for these materials are given in Table 1.
To determine the eect of temperature on paint or gesso applied
to dierent substrates, it is again possible to use Equation 2. Note that
changes in temperature will change the moisture content of materials
even when the ambient RH is held constant. At a constant RH, heating
will desiccate materials somewhat, and cooling will increase their moisture
content. The following discussion does not take these eects into account.
Figure 16 plots the calculated mechanical strains of ake white oil paint
directly applied to panels in the longitudinal, tangential, and radial direc-
tions of the wood, and to a copper panel as well. Because the thermal
coecient of expansion of the paint is greater than the thermal coecient
of wood in any direction, the paint responds to drops in temperature by
developing tensile strains. The woods shrinkage in the tangential and
radial directions relieves a considerable amount of the paint strain, since
the coecients in these directions more closely match those of the paint.
In the longitudinal direction of the wood, the coecient is the smallest
and strain relief to the paint the lowest. Hence, the greatest mechanical
strain increase in the paint occurs in the direction parallel to the grain of
the wood. As the temperature drops, the paint may pass through its glass-
transition temperature (T
g
). At this approximate temperature, the paint
undergoes a transition from ductile to very brittle and glassy. Below T
g
, the
paint is very fracture sensitive and prone to crack under low stresses and
strains. In this example, cracks could result when the strains reach levels
as low as 0.002. In the longitudinal direction of a wooden panel painting,
cracking occurs if the temperature drops from 22 C to approximately
19 C. A copper panel painting, however, requires a temperature drop
to 35 C to produce the same strain level.
40 20 0 20
Temperature (C)
M
e
c
h
a
n
i
c
a
l
s
t
r
a
i
n
Oaktangential
Oil paints
Oak
longitudinal
0.0035
0.0030
0.0025
0.0020
0.0015
0.0010
0.0005
0.0000
0.0005
Oakradial
31 Richard fig 16 eps
Copper
T
G
Fi gure 16
Calculated temperature-related strains in ake
white oil paint when applied to white oak and
copper. The paint strains in the longitudinal
direction are the highest, and failure can most
likely occur when the temperature drops
below the glass-transition temperature (T
g
).
This type of failure results in cracks in the oil
paint perpendicular to the grain of the wood.
Tabl e 1 Thermal coecients of expansion
of selected painting materials
Thermal coecient
Material of expansion
White oaklongitudinal 0.0000038/C
White oaktangential 0.0000385/C
White oakradial 0.00003/C
Oil paint 0.000052/C
Gesso 0.00002/C
Hide glue 0.000025/C
Copper 0.000017/C
Cracking in varnish and polyurethane coatings on wood has, in
fact, been recorded when the temperature has dropped from 24 C to
20 C. In the radial and tangential directions of the wood, the tempera-
ture must drop to well below 50 C to produce similar strains in the oil
paint layers.
It is unlikely that cracks in oil paint layers could occur perpen-
dicular to the grain of the wood because of RH variations. However,
with regard to temperature, even moderate subfreezing temperatures can
crack oil paint in this direction. Low temperatures are less likely to cause
cracking of paint parallel to the grain, unless the wooden support panel is
fully restrained from thermal movement during the temperature drop. As
Figure 16 shows, oil paint layers applied to copper can survive a substantial
drop in temperature. Note that resultant embrittlement of the paint layer
is far more severe when it is exposed to low temperature at moderate RH
than when exposed to low RH at room temperature.
Other paint media suer embrittlement similar to that suered
by oil paint, but at higher temperatures. With alkyd paints, a T
g
occurs at
approximately 5 C, while with acrylic paints, it occurs at approximately
5 C. While unlikely, it is possible for the temperature inside packing cases
to drop to 5 C in the cargo holds of aircraft, on the airport tarmac, or
inside an unheated truck. T
g
should be considered the lowest allowable
temperature for a safe environment, because embrittled materials are
more vulnerable to damage.
The eect of temperature on gesso applied to wooden panel
paintings is dierent from the eect of the same temperature on paint
applied to wooden panels. In general, gesso has a low thermal coecient
of expansion that is higher than that of the longitudinal direction of
white oak and lower than the oak coecients in the radial and tangential
directions. Figure 17 plots the calculated temperature-related mechanical
strains in the three dierent grain orientations for a gesso coating applied
to a white oak panel. First, the developed mechanical strains are minimal,
even at 40 C. In the longitudinal direction the gesso strains are tensile,
and in the tangential and radial directions they are compressive. Thus, it
542 Ri chard, Me ckl e nburg, and Tumos a
40 20 0 20
Temperature (C)
M
e
c
h
a
n
i
c
a
l
s
t
r
a
i
n
i
n
t
h
e
g
e
s
s
o
Compression response
Oaktangential
Tension response
Oaklongitudinal
0.0012
0.0010
0.0008
0.0006
0.0004
0.0002
0.0000
0.0002
0.0004
0.0006
0.0008
0.0010
0.0012
0.0014
Oakradial
31 Richard fig 17 eps
Fi gure 17
Calculated temperature-related strains in
gesso when applied to white oak. The gesso
strains in the longitudinal (tensile) and cross-
grain (compressive) directions are never very
high, and failure is not likely to occur, even if
the temperature drops signicantly.
543 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
appears that temperature has a signicantly smaller eect on gesso than
it has on oil paint.
In the panel itself, the most probable damage would occur in the
tangential direction if the wood were fully restrained and subjected to a
drop in temperature. The tangential direction has the highest thermal
coecient of expansion and the lowest strength. However, even in this
direction, a drop in temperature from 22 C to 40 C causes a mechani-
cal strain of only 0.00246, which is not a serious concern for wood.
Excessive heat can cause undue softening of paint and varnish
layers and therefore is to be avoided. In the transport environment, tem-
perature changes can be great enough to cause damage to the paint (and
varnish) layers. Thus, precautions must be taken to avoid exposing panel
paintings to extremes of hot or cold environments.
Temperature variations are inevitable in most transport situations
(Saunders 1991; Ostrem and Godshall 1979; Ostrem and Libovicz 1971).
Although variations are usually minimal during a local move in a climate-
controlled vehicle, they can grow extreme during a long truck trip during
harsh winter months. In the northern United States and Canada, for
example, winter lows of 20 C are typical, and temperatures of 40 C
are possible. These extremely low temperatures can cause damage to panel
paintings and must be avoided.
In the summer, temperatures of 4050 C can be found in many
parts of the world; because of solar heating, temperatures inside station-
ary vehicles can be even higher. High temperatures are less likely to cause
cracking in panel paintings, since heat softens the paint. However, var-
nishes can become tacky at high temperatures, causing wrapping materials
to adhere to the panel surface. The use of climate-controlled vehicles for
transporting works of art is the best way to minimize temperature varia-
tions, but contingency plans should be made in case of mechanical prob-
lems with vehicles or with their climate-control systems. Should a problem
occur, insulation in packing cases will slow the rate of temperature change
inside packing cases, but for only a short while (Richard 1991a).
Temperature variations can also occur in the cargo holds of air-
craft. Cargo holds of all modern commercial aircraft now have heating sys-
tems, however, and barring mechanical failure, the temperature should not
fall below 5 C. Acrylic paintings are at high risk at these lower tempera-
tures, but sound oil paintings on panel are not.
In addition to environmental variations, handling can add sucient stress
to a panel structure to cause paint loss, propagate cracks, separate joints,
and permanently deform its wood.
Shocks in the transport environment are derived from three basic
sources: handling before a work is packed, handling of the packing case, and
the motion of the vehicle carrying the packing case. Shock levels in trucks
and planes are low if packing cases are properly secured to the vehicle. In
contrast, handling operations are generally considered as imposing the
most severe loads on packages during shipment (Marcon 1991:123).
Packaging designers have achieved reasonable success in preventing ship-
ment losses due to shock by designing packages and cushioning systems
according to the presumption that shocks received during handling opera-
tions will be the most severe received by the packages during the entire
shipment (U.S. Department of Defense 1978:9).
Shock
544 Ri chard, Me ckl e nburg, and Tumos a
Because old panel paintings are fragile, the shock level to which
they are exposed must be minimized. The fragility factor, or G factor, is a
measure of the amount of force required to cause damage, and is usually
expressed in Gs. Mass-produced objects are destructively tested to measure
their fragility, but obviously this test is not possible with works of art.
Until recently, no attempt has been made to determine the fragility-factor
range for panel paintings. Instead, art packers have relied on estimates.
Conservatively, a packing case should ensure that a panel painting is not
subjected to an edge-drop shock level greater than 40 G. The edge drop,
however, is not the greatest concern.
One of the most serious accidents can occur when a painting
resting upright on the oor and leaning against a wall slides away and
falls to the oor. Another possible accident involves a case toppling over.
In both of these handling situations, a panel painting is at serious risk
because of inertially induced bending forces applied to the panel. The
bending stresses induced in a panel are potentially the most damaging,
and the thinner the panel, the greater the risk. While a thin panel has a
low weight (low mass), for a given action, the bending stresses increase
as a function of the inverse square of the thickness of the panel. For
example, consider a sound, 2.54 cm thick white oak panel painting mea-
suring 100 cm in the direction perpendicular to the grain, and 150 cm
in the direction parallel to the grain. If this panel painting is bowed and
supported in a frame, it is very likely that the support is along the two
long edges (Fig. 18). If this painting were to topple so that the rotation
were along one of the long edges, there would be bending stresses in the
wood perpendicular to the grain. These stresses can be calculated by rst
determining the eective loading on the panel that results at the time of
impact. If the impact were 50 G, the maximum bending stresses would
be approximately 4.66 Mpa. This stress is calculated by rst determining
the shear (Fig. 19) and bending (Fig. 20) resulting from the impact forces.
White oak has a specic gravity of approximately 0.62, which means
that it has a density of approximately 0.171 kg cm
3
. At 50 G, the density
of the wood is 0.032 kg cm
3
along the impact edge and diminishes to
zero at the rotating edge. For a 2.54 cm thick panel, the loading for
every 2.54 cm of width of the panel at the impact edge is 0.032 kg cm
3
,
and the loading tapers to zero at the other edge (Fig. 18). From the
bending moment diagram, the bending stresses can be calculated from
the equation
M
c
/ I (3)
where: is the bending stress, in either tension or compression, at the
outer surfaces of the panel; M is the bending moment calculated and
shown in Figure 20; c is one-half the thickness of the panel; I is the second
area moment of the cross section of the panel segment under considera-
tion, and I bd
3
/12, where b is the width of the panel section, and d is the
thickness of the panel.
The calculated bending stresses resulting from a 50-G topple
impact to a 100 150 2.54 cm thick oak panel are shown in Figure 21.
The maximum stresses are stationed approximately 58 cm from the rotat-
ing edge and reach 4.88 Mpa. This amount is slightly more than half the
breaking strength of structurally sound oak in the tangential direction.
50 G
31 Richard fig 18 eps

Approximate load
distribution resulting
from the topple
Station
Thickness of
the panel
Width of the panel (cm)
0 100
Fi gure 18
Approximate loading that occurs to a panel
painting subjected to a 50-G topple accident.
In this case, it is assumed that the panel is
supported only along the two parallel-to-grain
directions. It is always better to support the
panel continuously around the edges.
545 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
If the same event occurred to an oak panel 1.25 cm thick (with the
other two dimensions the same), the bending stresses would be 9.8 Mpa.
Even though the 1.25 cm thick panel weighs half as much as the 2.54 cm
one, it incurs twice the stress. The measured breaking stress of white oak
at room temperature and 50% RH is approximately 8.9 Mpa. The thinner
panel will likely crack in a 50-G topple accident. The 2.54 cm thick panel
would require a 100-G topple impact to crack it. If either panel were sup-
ported continuously around the edges, the risk of damage would decrease
by a factor of ve.
0 20 40 60 80 100
Station (cm)
S
h
e
a
r
f
o
r
c
e
(
N
)
2
1
0
1
2
3
4
31 Richard fig 19 eps
Fi gure 19
Shear in newtons (N) for a 2.54 cm wide strip
of a 100 150 x 2.54 cm thick panel subjected
to a 50-G topple accident.
0 20 40 60 80 100
Station (cm)
B
e
n
d
i
n
g
m
o
m
e
n
t
(
c
m
-
N
)
80
70
60
50
40
30
20
10
0
31 Richard fig 20 eps
Fi gure 20
Bending-moment diagram for a 2.54 cm wide
strip of a 100 150 2.54 cm thick panel
subjected to a 50-G topple accident. The
bending moments of panels subjected to
topples can be quite high.
546 Ri chard, Me ckl e nburg, and Tumos a
Figure 22 shows the calculated bending stresses of oak panels of
dierent sizes and thicknesses subjected to 50-G topple impacts. These
panels are assumed to be supported on the parallel-to-grain edges only,
and the topple is a rotation of one of those edges. For this test, it is also
assumed that there are no battens or cradles attached to the reverse, since
they would provide a certain degree of bending protection.
Panels constructed of lighter woods such as pine (Pinus spp.;
specic gravity, 0.34) will develop comparatively lower bending stresses
when subjected to a 50-G topple impact. However, the strength of the
0 20 40 60 80 100
Station (cm)
B
e
n
d
i
n
g
s
t
r
e
s
s
e
s
(
M
P
a
)
6
5
4
3
2
1
0
31 Richard fig 21 eps
Fi gure 21
Distribution of the calculated bending stresses
for a 2.54 cm wide strip of a 100 150 2.54
cm thick panel subjected to a 50-G topple
accident. The bending stresses of panels sub-
jected to topples can be quite high, and in this
case they reach about one-half the breaking
stress of oak in the tangential direction.
Thinner panels are at even greater risk.
0 20 40 60 80 100 120
Cross-grained width (cm)
M
a
x
i
m
u
m
b
e
n
d
i
n
g
s
t
r
e
s
s
(
M
P
a
)
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
31 Richard fig 22 eps

1.25 cm thick
2.5 cm thick
1.9 cm thick
Breaking strength
Fi gure 22
Calculated maximum bending stresses for
white oak panels of dierent thicknesses and
sizes when subjected to 50-G topple accidents.
These stresses assume that the panels are sup-
ported only on the two parallel-to-grain edges.
lighter wood is also lower, and the result is that the risk of damage is greater
than for denser woods. Figure 23 illustrates the results of the calculated
bending stresses for dierent thicknesses of oak and pine panels of 100
150 cm subjected to 50-G topple impacts. The breaking stress of the pine in
the tangential direction is only 3.10 Mpa. As was the case with white oak,
the thinner pine panels are at greater risk, and the pine panels must be
thicker than oak panels to prevent failure under the same topple conditions.
This implies that a single packing criterion is not sucient for the
impact protection of panel paintings. Larger and thinner panel paintings
obviously need greater protection than those that are smaller and thicker.
In addition, in this analysis it is assumed that the panel is sound, since
existing cracks reduce the total strength. Panel paintings should be sup-
ported continuously around the edges in a way that allows them to expand
and contract with RH and thermal uctuations. Special care should be
taken to prevent topple accidents; one way to do this is to pack more than
one painting in a case, eectively increasing the width of the case and
reducing the possibility of a topple.
Panel paintings in the size range of 100 150 cm will often be
thicker than 2.54 cm, and those that are thinner are probably supported by
either battens or cradles. Yet a 2.54 cm thick oak panel that is 125 cm wide
or greater will fail in a 50-G topple. Based on this information, a 30-G
maximum impact criterion for topple should be considered reasonable.
It should not be dicult to provide 30-G topple protection for
larger panels. For one thing, the risk for an edge drop is much lower. It is
fairly easy to provide 40-G protection for edge drop heights of 75 cm or
less, using foam cushioning materials (the use of foam cushioning to
reduce shock will be discussed below).
The primary sources of vibration in the transit environment come from
the vehicles used for transport. Trucks impose the severest vibration
loads on cargo with the railcar next, followed by the ship and aircraft
Vibration
547 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6
Panel thickness (cm)
M
a
x
i
m
u
m
b
e
n
d
i
n
g
s
t
r
e
s
s
(
M
P
a
)
12
11
10
9
8
7
6
5
4
3
2
1
0
31 Richard fig 23 eps

White oak
Pine breaking strength
Clear white pine
Oak breaking strength
Fi gure 23
Calculated maximum bending stresses for
100 150 cm white oak and pine panels sub-
jected to 50-G topple accidents versus panel
thickness. These stresses assume that the pan-
els are supported only on the two parallel-to-
grain edges. Even though the pine is a lighter
wood, because of its substantially lower
strength, panels made from it are at serious
risk in the event of a topple.
(Ostrem and Godshall 1979:29). In trucks, the main sources of vibration
are the natural frequencies of its body, engine, tires, drive train, and sus-
pension system. The properties of the road surface are also a factor. The
vibration levels in vehicles are all relatively low and random in nature, as
vehicles are usually designed for passenger comfort.
Low levels of vibration are unlikely to damage panel paintings
unless sustained vibrations create resonant vibrations in the panel; the ran-
dom nature of vehicle vibration makes this unlikely. In addition, the reso-
nant frequencies of panel paintings are high, and those vibrations are
easily attenuated by packing cases (Marcon 1991:112).
There are many packing-case designs suggested for the transport of panel
paintings. It is essential that all cases provide adequate protection against
shock, vibration, and environmental uctuations. Protection against the
rst two stresses is usually achieved through the use of foam cushioning
materials. Although various cushioning materials are available for the
transport of works of art, the most commonly used are polyethylene and
polyester urethane foams. These foam products, along with polystyrene
foam, can additionally function as thermal insulation. The proper use of
these materials and information concerning the principles of case design
are available in many publications (Mecklenburg 1991; Piechota and
Hansen 1982; Richard, Mecklenburg, and Merrill 1991; Stolow 1966, 1979,
1987) and will only be summarized here.
Packing-case construction
Packing cases for panel paintings should be rigid to ensure that panels do
not ex or twist during handling and transport. Rigidity can be accom-
plished by the use of relatively sti materials and quality construction
techniques. It is recommended that glue be used in the joinery of the cases
because it increases the strength and stiness of the joints. Case joints held
together with only nails or screws perform poorly when dropped. A case
having edges and corners that are well-joined can have over ten times the
strength and one hundred times the rigidity of a case that has corners and
edges that are poorly joined (Richard, Mecklenburg, and Merrill 1991).
Compared to single packing-case designs, double packing cases
provide signicantly better protection for panel paintings because an inner
case adds rigidity to the structure. An inner case also increases the level of
thermal insulation and reduces the likelihood of damage should the outer
case be punctured by a sharp object, such as the blade of a forklift.
Figure 24 depicts a double packing-case design commonly used at
the National Gallery of Art in Washington, D.C. The polyester urethane
foam not only functions as a cushioning material but also provides thermal
insulation. The entire case is lined with a minimum of 5 cm of foam, which
proves adequate insulation for most transport situations if temperature-
controlled vehicles are used. A packing case for a typical easel-sized painting
has a thermal half-time of two to three hours (Fig. 25) (Richard 1991a). The
foam thickness should be increased to at least 10 cm if extreme temperature
variations are anticipated. However, thermal insulation only slows the rate
of temperature change within the case: increasing the thickness of the insu-
lation increases the thermal half-time to approximately four to ve hours.
Packing Cases
548 Ri chard, Me ckl e nburg, and Tumos a
When paintings are transported in extreme climates, the only way to main-
tain temperature levels that will not damage paintings is through the use of
temperature-controlled vehicles.
Foam-cushion design
In the packing-case design depicted in Figure 24, the polyester urethane
foam provides shock protection for the painting. The painting should be
rmly secured within the inner case. There are two procedures that are
commonly used: (1) the paintings frame is secured to the inner case with
metal plates and screws, or (2) the frame is held in place with strips of
foam. Shock protection in a double case design is provided by foam cush-
ions tted between the inner and outer cases. When a packing case is
dropped, the foam cushions compress on impact, allowing the inner case
549 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
Polyester
urethane foam
31 Richard fig 24 eps

Polyester urethane foam

Polyester
urethane foam
corner pads
Polyester urethane foam corner pads
Fi gure 24
Interior view of the outer packing case of a
double-case packing system.
0 4 8 12 16 20 24 28
Time (hours)
T
e
m
p
e
r
a
t
u
r
e
(

C
)
20
15
10
5
0
31 Richard fig 25 eps

A
5 2 1
Fi gure 25
Thermal half-times for three dierent case
designs. The cases were initially conditioned
at 20 C environment. The half-times shown
are 1, 2, and 5 hours. Even for a well-insulated
case (such as the 5-hour half-time example),
little time is available before the case equili-
brates to a new temperature.
to move within the outer case. While the acceleration of the outer case is
quickly halted on impact with the oor, the acceleration of the inner case
is halted much more slowly. If the packing system functions properly, the
outer case may sustain a few hundred Gs on impact, while fewer than 50 G
are transmitted to the inner case and the painting inside.
It is easy to attain 50-G protection for panel paintings when pack-
ing cases are dropped less than 1 m. In fact, if careful attention is given
to the proper use of foam cushioning materials, 25-G protection can be
attained. The shock-absorbing properties of cushioning materials are pro-
vided in graphs known as dynamic cushioning curves (Fig. 26). These curves
plot the G forces transmitted to a packed object as a function of the static
load of the cushioning material. The curves vary with dierent materials,
thicknesses, and drop heights. Dynamic cushioning curves for many
materials are published in the Military Standardization Handbook (U.S.
Department of Defense 1978). More accurate cushioning curves for
specic products are usually available from the manufacturers. The use
of these curves has been extensively discussed in several publications
(Piechota and Hansen 1982; Richard 1991b).
Two cushioning curves for polyester urethane foam with a density
of 33 kg m
3
are shown in Figure 26. Both are calculated for a drop height
of 75 cm. Note that an increase in foam thickness dramatically eects the
cushioning properties of the material. The lowest point on each curve cor-
responds to the optimal performance for a given thickness of the material.
Therefore, as seen in Figure 26, the optimal static load for 10 cm thick
polyester urethane foam is approximately 0.025 kg cm
2
(point A, Fig. 26).
The static load is the weight of the object divided by the area in contact
with the foam cushioning. At this static load, a painting packed with
10 cm thick cushions of polyester urethane foam will sustain a shock force
550 Ri chard, Me ckl e nburg, and Tumos a
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14
Static load (kg cm
2
)
P
e
a
k
a
c
c
e
l
e
r
a
t
i
o
n
(
G
)
120
110
100
90
80
70
60
50
40
30
20
10
0
31 Richard fig 26 eps

A
10-cm polyester
urethane foam
B
5-cm polyester
urethane foam
Fi gure 26
Dynamic cushioning curves for two thick-
nesses of polyester urethane foam. The curves
show the distinct advantage of using the
thicker material.
of approximately 22 G. If the cushioning in the packing case is 5 cm thick,
then the optimal static load is 0.016 kg cm
2
and a force of 45 G would be
anticipated (point B, Fig. 26). Because of the dramatic improvement in the
performance of the 10 cm thick foam as compared to the 5 cm thick foam,
it is highly recommended that foam cushions at least 10 cm thick be used
in packing cases built for the transport of panel paintings.
It is not possible to predict the fragility of every panel painting
accurately, although the methods described provide a good estimate for
reasonably sound objects. Due to cracks and unseen defects, panel paintings
will always be morenever lessfragile than calculated. Manufacturing
companies that sell mass-produced items destructively test a few to ascer-
tain their fragility. In this way, the company can design an adequately
protective package at the least possible cost. While a small percentage of
the items will be damaged, the expense incurred due to loss will be less
than the cost of more complex and expensive packing cases. In the absence
of accurate fragility information, it is recommended that packing cases
provide at least 40-G protection for small panel paintings and 30-G pro-
tection for larger panel paintings. To provide optimal performance, the
foam cushions should be at least 10 cm thick, and the static load on the
foam should be calculated, using dynamic cushioning curves, to provide
optimal performance.
Wrapping paintings in moisture-barrier materials is one way to control
their moisture content during transport (Hackney 1987). Relatively thick
polyethylene lms that are well sealed with packaging tape usually work
eectively. The quality of commercial polyethylene lm materials varies
considerably, however: the lm is often made from recycled materials, and
a low-quality lm might result from the addition of grease, oil, chemical
additives, and powders during the manufacturing process. Better moisture-
barrier materials are available, but in ordinary transport situations, they
provide few advantages over polyethylene sheeting, provided it is of high
quality. It would be advantageous, however, to use the better materials
when paintings are stored for many weeks in an environment having
extremely high or low RH, or one having high concentrations of atmos-
pheric pollutants.
Conservators and packers are often concerned that wrapping
paintings in a moisture barrier causes condensation. Condensation prob-
lems can occur in packing cases containing large volumes of air relative to
the mass and surface area of the hygroscopic materials inside. However,
when a typical panel painting is wrapped in polyethylene, the volume of
air is very small relative to the mass and surface area of the painting and
frame. In this case, experimental evidence indicates that condensation will
not occur unless a painting is acclimated to a very high RH level (at least
70%) and is exposed to a rapid and extreme temperature drop in a nonin-
sulated packing case. The most likely cause of condensation is unpacking
and unwrapping a cold painting in a warm room (those who wear eye-
glasses have experienced similar condensation problems when they walk
indoors on a cold winter day). This problem can be avoided simply by
allowing several hours for the painting to acclimate to the higher tempera-
ture while it is still in the insulated case.
Wrapping paintings in polyethylene or an alternate moisture-
barrier material is particularly important when there is uncertainty about
Wrapping Materials for
Paintings
551 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
the environment in which the packing cases will be stored. Most packing
cases contain hygroscopic materials, and if they are stored in environments
having an unusually high or low RH, they acclimate to that environment.
Unless sucient time (usually a week or two) is allowed for the cases to
reacclimate to the proper RH before packing, inappropriate microenviron-
ments may be created in the cases. Similar problems can occur when pack-
ing cases are constructed from wood that has not been acclimated to the
proper RH; a moisture-barrier lm surrounding the painting reduces the
potential of damage from an inappropriate environment.
To improve the microclimate inside packing cases, buering mate-
rials such as silica gel can be added. Additional buering materials slow the
variation of moisture content in the painting, should it be subjected to
extreme variations of RH for an extended period of time. The greatest risk
in adding silica gel to a packing case is the possibility of using improperly
conditioned silica gel. Even if the gel is carefully conditioned by the lend-
ing institution, it is always possible that it has become improperly condi-
tioned during the period when the packing cases were in storage. Therefore,
if silica gel is used, it is essential that it be checked for proper conditioning
each time it is packed.
Silica gel can also be used in a microclimate display case in which
the painting remains during exhibition. A properly constructed display
case provides a stable microclimate environment for a panel painting and
is particularly useful when a painting is accustomed to an environment
that the borrowing institution cannot achieve. A panel acclimated to 65%
RH, for example, could be placed in a microclimate display case while on
loan to a borrowing institution that can only maintain 35% RH during
winter. It must be kept in mind, however, that mold growth can develop
inside microclimate display cases acclimated to a high RH.
Because of concerns about their fragility, panel paintings are often hand
carried by courier during transit. In certain situations, there are advantages
to hand carrying works of art. The work remains in the possession of the
courier at all timesa situation not possible if works are sent as cargo on
an aircraft. The painting will be subjected to smaller temperature varia-
tions if the courier is conscientious about time spent in unusually cold or
warm locations. However, there are some risks associated with hand carry-
ing works of art. It is important that the painting t into a lightweight but
sturdy case that is easily carried and small enough to t in a safe location
on an aircraft, ideally under the seat. Overhead compartments should not
be used because the work could accidentally fall to the oor should the
compartment door open during the ight. The case might be placed in an
aircraft coat closet if necessary, but it must be secured so that no move-
ment can occur.
Another risk with hand carrying works of art is theft. Carried
materials of high value are a potential target for well-informed thieves.
Although this is an extremely rare problem, it is a concern that neverthe-
less must be considered. While couriers may feel more secure because
they are never separated from their packing cases, this proximity doesnt
necessarily mean that the work is actually safer.
There are many ways to pack a panel painting for hand carrying on
an aircraft. Metal photographic equipment cases have proved very success-
Hand Carrying Panel
Paintings
552 Ri chard, Me ckl e nburg, and Tumos a
553 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
ful. These cases come in various sizes and shapes, the smaller ones tting
conveniently under aircraft seats. The procedure for packing a painting in
these cases is straightforward. The National Gallery of Art in Washington,
D.C., often follows these steps: First, either the framed panel painting is
wrapped directly in polyethylene and sealed with waterproof tape, or it is
placed in an inner case that is wrapped in polyethylene. Unframed panels
are always tted into an inner case to prevent anything from touching the
surface of the painting. The metal photography case is then lled with
polyester urethane foam. A cavity is cut into the foam with a minimum of
2.54 cm of foam remaining on all sides. In this cavity, the wrapped painting
or inner case is placed. In this procedure the polyester urethane foam func-
tions as both cushioning material and thermal insulation.
Most panel paintings that are in good condition and free to respond
dimensionally to environmental variations can be safely transported, as
long as they are packed properly. However, there are circumstances when
some paintings are at greater risk than others. Therefore, all panels should
be carefully examined and an assessment should be made of RH- and
temperature-related stresses that may develop from improper framing
techniques or from restraint imposed by cradles or battens. Existing
cracks in the design layers usually act as expansion joints, but cracks in
panels can prove to be a potential problem, especially if the painting is
subjected to impact.
It is also important to compare the RH levels where the painting
normally hangs to the RH levels at the borrowing institution. If there is a
large discrepancy in the RH, a microclimate display case could be used.
Tables 24 summarize the relative RH-related risks for sample paintings of
dierent construction and grain orientation. For example, Table 2 shows
the risks of transporting a restrained, tangentially cut, white oak panel
that has been equilibrated to 70% RH or higher.
Tables 3 and 4 show that it is potentially hazardous to ship a panel
painting that has been equilibrated to 70% RH or higher and that has a
gesso ground or paint directly applied to the woodparticularly if the
wooden support is tangentially cut and not restrained.
To maintain stable moisture contents, paintings should be
wrapped in moisture-barrier materials, provided they are not already con-
ditioned to an unusually damp environment. Because condensation can
occur when paintings acclimated to very high RH are transported in
extremely cold weather, such transport could encourage mold growth.
Conclusion
Tabl e 2 Maximum allowable RH ranges and relative risks for sound, uncracked, and restrained
white oak panels in dierent grain orientations
Panel grain Equilibrium RH Allowable RH
orientation (%) range to yield (%) Relative risk
Tangential 36 2554 medium
Tangential 50 3363 low
Tangential 70 6273 high
Radial 50 2375 low
Radial 70 4085 low
554 Ri chard, Me ckl e nburg, and Tumos a
Temperature variations during transit should be minimized by use
of climate-controlled vehicles and thermal insulation inside packing cases.
Table 5 gives the typical glass-transition temperatures for three types of
paint. However, paintings should never be subjected to temperatures as
low as these values and, ideally, should stay above 10 C.
Careful attention should be given to the selection and proper use
of cushioning materials in the packing cases to ensure that paintings are not
exposed to edge drops resulting in forces exceeding approximately 4050 G.
For panel paintings, topple accidents can cause more severe dam-
age than edge drops. The edges of panel paintings should be supported
Tabl e 3 Maximum allowable RH ranges and relative risks for well-attached gesso applied to
unrestrained white oak panels in dierent grain orientations
Panel grain Equilibrium RH Allowable RH
orientation (%) range to yield (%) Relative risk
Longitudinal 50 2086 low
Radial 50 2279 low
Tangential 50 3362 medium
Longitudinal 64 2993 low
Radial 64 3387 low
Tangential 64 5368 high
Longitudinal 70 3296 low
Radial 70 3284 low
Tangential 70 6573 very high
Longitudinal 36 1275 low
Radial 36 1571 low
Tangential 36 2654 medium
Tabl e 4 Maximum allowable RH ranges and relative risks for well-attached oil paint applied to
unrestrained white oak panels in dierent grain orientations
Panel grain Equilibrium RH Allowable RH
orientation (%) range to yield (%) Relative risk
Longitudinal 50 895 low
Radial 50 1386 low
Tangential 50 2765 medium
Longitudinal 64 1695 low
Radial 64 2092 low
Tangential 64 4371 medium
Longitudinal 70 1795 low
Radial 70 1990 low
Tangential 70 6175 very high
Longitudinal 36 492 low
Radial 36 888 low
Tangential 36 2260 medium
Tabl e 5 Approximate glass-transition tem-
peratures for selected paints
Glass-transition
Material temperature, T
g
(C)
Oil paint 10
Alkyd paint 5
Acrylic paint 5
continuously around the edges when in the frame and during transport.
The panel must be free to move in response to changes in temperature
and RH. See Table 6 for the approximate topple-accident G levels that
will break uncracked panels of various dimensions and woods. This table
assumes that there is no auxiliary support, such as battens or cradles
attached to the panels, and that the wood is cut in the tangential direction.
Woods cut in the radial direction are approximately 40% stronger than the
examples provided in Table 6.
Low temperatures can severely reduce the eectiveness of foam
cushions in reducing impact G levels.
Normally, transit vibration in panel paintings can be successfully
attenuated by the foam cushions used to protect the painting from
impact damage.
Hackney, S.
1987 The dimensional stability of paintings in transit. In ICOM Committee for Conservation 8th
Triennial Meeting, Sydney, Australia, 611 September 1987, Preprints, 597600. Marina del
Rey, Calif.: Getty Conservation Institute.
Marcon, P.
1991 Shock, vibration, and the shipping environment. In Art in Transit: Studies in the
Transport of Paintings, ed. M. F. Mecklenburg, 12132. Washington, D.C.: National
Gallery of Art.
Mecklenburg, M. F., ed.
1991 Art in Transit: Studies in the Transport of Paintings. Washington, D.C.: National
Gallery of Art.
Mecklenburg, M. F., and C. S. Tumosa
1991 Mechanical behavior of paintings subjected to changes in temperature and relative
humidity. In Art in Transit: Studies in the Transport of Paintings, ed. M. F. Mecklenburg,
173216. Washington, D.C.: National Gallery of Art.
Mecklenburg, M. F., C. S. Tumosa, and D. Erhardt
1998 Structural response of painted wood surfaces to changes in ambient relative humidity.
In Painted Wood: History and Conservation, ed. Valerie Dorge and F. Carey Howlett. Los
Angeles: Getty Conservation Institute.
References
555 Ttcni cai Cos i ntrari os ror rnt Tras rorr or Pati Pai ri os
Tabl e 6 Topple-accident G levels required to break selected wooden panels cut in the tangential
direction and supported along the two parallel-to-grain directions
Panel width Panel thickness Topple G at failure: Topple G at failure:
(cm) (cm) White oak Pine
127 1.25 29 19
127 1.90 44 28
127 2.53 59 37
102 1.25 46 29
102 1.90 69 44
102 2.53 92 58
76 1.90 82 52
76 1.90 122 77
76 2.53 163 103
Mecklenburg, M. F., C. S. Tumosa, and M. H. McCormick-Goodhart
1995 A general model relating applied forces to environmentally induced stresses in
materials. Materials Issues in Art and Archaeology, vol. 352, ed. P. B. Vandiver,
J. R. Druzik, J. L. Galvan Madrid, I. C. Freestone, and G. S. Wheeler, 28592.
Pittsburgh: Materials Research Society.
Ostrem, F. E., and W. D. Godshall
1979 An assessment of the common carrier shipping environment. USDA General Technical
Report No. 22. Madison, Wisc.: Forest Products Laboratory.
Ostrem, F. E., and B. Libovicz
1971 A survey of environmental conditions incident to the transportation of materials.
GARD Report No. 15121. Niles, Ill.: General American Research Division.
Piechota, D., and G. Hansen
1982 The care of cultural property in transit: A case design for traveling exhibitions.
Technology and Conservation 7(4):3246.
Richard, M.
1991a Control of temperature and relative humidity in packing cases. In Art in Transit:
Studies in the Transport of Paintings, ed. M. F. Mecklenburg, 27997. Washington, D.C.:
National Gallery of Art.
1991b Foam cushioning materials: Techniques for their proper use. In Art in Transit: Studies in
the Transport of Paintings, ed. M. F. Mecklenburg, 26978. Washington, D.C.: National
Gallery of Art.
Richard, M., M. F. Mecklenburg, and R. Merrill
1991 Art in Transit: Handbook for Packing and Transporting Paintings, 14. Washington, D.C.:
National Gallery of Art.
Saunders, D.
1991 Temperature and relative humidity conditions encountered in transportation. In
Art in Transit: Studies in the Transport of Paintings, ed. M. F. Mecklenburg, 299309.
Washington, D.C.: National Gallery of Art.
Stolow, N.
1966 Controlled Environments for Works of Art in Transit. London: Butterworths.
1979 Conservation Standards for Works of Art in Transit and on Exhibition. Paris: Unesco.
1987 Conservation and Exhibition: Packing, Transport, Storage, and Environmental Considerations.
London: Butterworths.
U.S. Department of Agriculture
1987 Wood Handbook: Wood as an Engineering Material. Agriculture handbook no. 72.
Madison, Wisc.: U.S. Forest Products Laboratory.
U.S. Department of Defense
1978 Military Standardization Handbook: Package Cushioning Design. U.S. Department
of Defense handbook no. MIL-HDBK-304B. Washington, D.C.: U.S.
Department of Defense.
Weaver, W., and J. M. Gere
1965 Matrix Analysis of Framed Structures. 2d ed. New York: Van Nostrand.
556 Ri chard, Me ckl e nburg, and Tumos a
557
Odile Ba entered the paintings conservation department of the Louvre Museum to work on the
Campana Collection in 1967; she was later appointed director of the department, which by then
served all of the museums in France. She collaborated with the cabinetmaker Ren Perche until
his retirement and then with his successor, Daniel Jaunard of the Atelier Claude Huot, until her
retirement in 1989.
Sgolne Bergeon became a professor of physics in 1965 and a museum conservator in 1969. She
joined the Service de Restauration des Peintures des Muses Nationaux in 1970, serving as direc-
tor from 1981 to 1988. Her career has included work at the Villa Medicis and the International
Centre for the Study of the Preservation and the Restoration of Cultural Property, Rome
(ICCROM), and she became the president of the ICCROM Council in 1993. She is the author of
several publications and has curated several conservation exhibitions.
George Bisacca is a conservator at the Metropolitan Museum of Art in New York, where he has
worked since 1983, and holds an adjunct professorship at the Institute of Fine Arts, New York
University. He trained in paintings conservation at the Palazzo Pitti with Andrea Rothe and Alo
Del Serra and specialized in the treatment of panel paintings with Renzo Turchi and Giovanni
Marrusich of the Opicio delle Pietre Dure.
Robert A. Blanchette, Ph.D., is a professor in the Department of Plant Pathology at the
University of Minnesota at Minneapolis St. Paul. He has written numerous scientic articles and
reviews on degradation processes of living trees and wood products and has coauthored two
books dealing with microbial degradation of wood. His current research involves biodeterioration
of archaeological wood from terrestrial and aquatic environments, as well as the development of
conservation methods for decayed wood.
Simon Bobak, conservator of paintings, currently works in London. He is a fellow of the
International Institute for Conservation. He also holds the position of honorary chief conservator
at the Hamilton Kerr Institute, Cambridge, England.
David Bomford, who is now senior restorer of paintings at the National Gallery, London, joined
that institution as a junior restorer in 1968, after postgraduate research in chemistry; he was
trained by Helmut Ruhemann. At the National Gallery, Bomford has worked on a large number
of paintings and has lectured and published widely, especially on the study of European painting
techniques. He was coorganizer and coauthor of the award-winning series of exhibitions and cata-
logues called Art in the Making, on the subjects of Rembrandt, fourteenth-century Italian paint-
ing, and Impressionism, and he also served for ten years as editor of the international journal
Studies in Conservation. In addition, Bomford serves as secretary-general of the International
Institute for Conservation. In 199697, he was the rst conservator to become Slade Professor
of Fine Art at Oxford University.
Jacqueline Bret is an engineer and a physicist at the Institut de Physique et de Chimie Industrielles
de Lyon. She holds an advanced degree from the Ecole du Louvre and is a research specialist with
the Service de Restauration des Muses de France, Petite Ecurie du Roy, Versailles.
Contributors
Al Brewer, a Canadian who learned much from his father with regard to forests, wood, and wood-
craft, received a B.Sc. in forestry from the University of New Brunswick, in eastern Canada (1983),
attended the Pennsylvania Academy of the Fine Arts for two years (197880), and received a
masters degree in art conservation from Queens University in Canada in 1987. Since then he
has conserved easel paintings, specializing in panel structural work, at the Hamilton Kerr Institute,
University of Cambridge, where he has also taught. More recently, he has concentrated on
researching the eects of overall reinforcement structures on the preservation of panel paintings.
Ciro Castelli began work as a joiner in 1957, progressing to the position of cabinetmaker for
a private company. In 1966 he began restoring panel paintings and wooden structures at the
Fortezza da Bassos state-run laboratory, including paintings damaged in the ood of 1966. Now
with the Opicio delle Pietre Dure e Laboratori di Restauro, Florence, Italywhere he has also
been a teacher since 1978he has restored important works by Masaccio, Giovanni del Biondo,
Raaello Sanzio, and Botticelli, among many others. As a consultant and restoration expert repre-
senting the public museums of Italy, Castelli has served on ocial delegations at international art
meetings. His reports have appeared often in OPD Restauro, as well as in restoration catalogues
and conservation congress transcripts.
Vinod Daniel received his M.Tech degree in chemical engineering in 1986 from the Indian
Institute of Technology in Madras, India, where he worked on the rheological characteristics of
polymer blends. He received his M.S. degree in physical chemistry in 1991 from Texas Christian
University, where his thesis addressed diusion in liquids. From 1991 to 1994 he was a senior
research fellow in the environmental sciences division at the Getty Conservation Institute. His
research involves museum cases, moisture buers in display cases, nontoxic fumigation, and data
acquisition. He is presently scientic ocer at the Australian Museum in Sydney, Australia.
Gilberte Emile-Mle completed advanced studies in art history at the Sorbonne. In 1950 she
became head of paintings restoration at the Louvre Museum and served as head conservator of
the Service de Restauration des Peintures des Muses Nationaux from 1971 until her retirement in
1981. She is the author of numerous articles on the history of the restoration of paintings.
Jean-Albert Glatigny, an art restorer, specializes in the treatment of wood supports. After study-
ing cabinetmaking, he took four years training at the Institut Royal du Patrimoine Artistique
(IRPA), Brussels, in the polychrome sculpture and panel painting workshops. In addition to his
restorers activities at IRPA and abroad, he teaches at several restoration schools and participates
in studies of works of art and conducts research on ancient techniques of woodworking.
Gordon Hanlon received his B.A. degree in biology from the University of York, England, in
1979. From 1980 to 1984 he was assistant curator of Road Transport and Agricultural Implements
at the Museum of Science and Technology, London. In 1984 he started a four-year studentship at
the Victoria and Albert Museum, London, specializing in the conservation of furniture and gilded
objects. In 1988 he joined the J. Paul Getty Museum as an intern and is now associate conservator
in the Decorative Arts and Sculpture Conservation department. He specializes in the conservation
of gilded furniture.
R. Bruce Hoadley holds a B.S. in forestry from the University of Connecticut, as well as
masters and doctorate degrees in wood technology from Yale University. He is currently a
professor in wood science and technology at the University of Massachusetts at Amherst,
where his principal teaching and research interests are the anatomy and fundamental properties
of wood. His wood identication analyses have been included in catalogues of major collec-
tions, including those of the Metropolitan Museum of Art in New York, the Garvan Collection
at Yale University, the furniture collection at the J. Paul Getty Museum, and the diplomatic
reception rooms at the U.S. Department of State. He is the author of two books, Understanding
Wood (1980) and Identifying Wood (1990), and of more than fty scientic and popular articles
relating to wood.
James S. Horns studied with Richard Buck from 1971 to 1974 at the Intermuseum Conservation
Association at Oberlin College, in Ohio, where he received a master of arts degree in conserva-
tion. He was a paintings conservator at the Minneapolis Institute of Arts from 1974 to 1978 and at
the Upper Midwest Conservation Association from 1979 to 1986. Since that time he has been a
conservator in private practice in Minneapolis.
558
559
Claude Huot carries on a two-generation tradition of cabinetmaking, having practiced under his
father, Georges Huot, whom he succeeded as director of his studio in 1962, and Ren Perche. An
avid personal interest in airplanes and gliders has allowed him to become better acquainted with
various uses of wood and its mechanical behavior, from the standpoints of both piloting and con-
serving aircraft made of wood and canvas.
Daniel Jaunard is a cabinetmaker and restorer of support frames for easel painting. He has previ-
ously worked with Ren Perche at the Atelier Claude Huot and is licensed by the Service de
Restauration des Muses de France.
Peter Klein graduated with a degree in wood technology from the University of Hamburg in
1973 and received a doctoral degree, with a specialty in wood science, from the same university in
1976. From 1976 to 1978 he served on the sta of the universitys Department of Wood Biology,
and from 1979 to 1981 he was a visiting scientist at the Gemldegalerie Berlin-Dahlem. Since 1981
he has served on the sta of the University of Hamburgs Department of Wood Biology. His
research activities concentrate on wood biology and technology, wood conservation and preserva-
tion, and dendrochronology.
Frdric J. M. Lebas studied at the Institut Suisse pour lEtude de lArt in Zurich; he later served
as paintings and sculpture restoration assistant to Th. Brachert at the Germanisches National
Museum, Nuremberg, and held a one-year fellowship to the Institut Royal du Patrimoine
Artistique de Bruxelles. Since 1979 he has served as restorer of paintings and sculpture at the
Museum fr Kunst und Gewerbe, Hamburg.
Patrick Mandron is a graduate of the Institut Franais de Restauration des Oeuvres dArt. A cabi-
netmaker and restorer of support frames for easel painting, he teaches at the Sorbonne, Universit
de Paris I, in the Matrise des Sciences et Techniques de la Conservation des Biens Culturels. He is
licensed by the Service de Restauration des Muses de France.
Raymond Marchant has a background in design engineering, carpentry, cabinetmaking, and fur-
niture restoration. He has also worked for John Bull in London as a technician restoring metal
sculpture. In 1989 he joined Simon Bobak in association with the Hamilton Kerr Institute (HKI),
Cambridge, England. At the London studio of the HKI, he works on the structural conservation
of panel paintings, and at the HKI in Whittlesford, Cambridge, he advises on the structural con-
servation of panel paintings.
Giovanni Marussich, who was born in Croatia, embarked on his professional life as a wood-
worker in 1948, and he immigrated to Florence in 1956. From 1962 to 1983 he was a wood con-
servator for panel paintings at the Fortezza da Basso, part of the Opicio delle Pietre Dure e
Laboratori di Restauro (formerly the Soprintendenza alle Gallerie) in Florence. Since then he has
done conservation on panel paintings for various institutions and has taught a course on wood
conservation at the Museo de Arte de Catalunya in Barcelona. He has also been a consultant to
the J. Paul Getty Museum. He is presently involved in a restoration campaign for war-ravaged
paintings in the former Yugoslavia.
Ian McClure studied English literature at Bristol University and art history at Edinburgh. He
became head of paintings conservation at Glasgow Art Gallery and Museum in 1978. In 1982 he
was named assistant to the director at the Hamilton Kerr Institute, where he became director in
1983. He has written on various specic conservation projects and techniques, as well as on the
history of conservation.
Marion Mecklenburg holds B.S., M.S., and Ph.D. degrees in structural engineering from the
University of Maryland. He has worked for twenty years as a paintings conservator in the United
States. In 1987 he joined the Conservation Analytical Laboratory of the Smithsonian Institution,
where he is a senior research scientist and where for several years he was the assistant director for
conservation research. He has also been an adjunct professor in the Department of Conservation
at the University of Delaware, an assistant professor and director of the Fracture Mechanics
Laboratory at the University of Maryland, and coordinator of the graduate program for material
science at the Johns Hopkins University. His research interests include the mechanics of materials
and the eects of the environment on the mechanical properties of materials.
Anthony M. Reeve is senior restorer at the National Gallery, London, where he joined the con-
servation department in 1963 and trained under Arthur Lucas, Helmut Ruhemann, and Louis
Howard. He carries out all forms of conservation, cleaning, restoring, and structural work on all
the paintings he conserves. Since 1977 he has been in charge of all structural work, research,
development, and application of improved methods of conservation. He represents the fourth
generation of picture restorers in his family.
Mervin Richard studied paintings conservation at Intermuseum Laboratory in Oberlin, Ohio, and
joined the sta of the laboratory upon graduation. He then held positions as a paintings conserva-
tor at the Philadelphia Museum of Art and at the Winterthur Museum. Since 1984 he has been
the head of exhibition conservation at the National Gallery of Art in Washington, D.C., where he
is also the deputy chief of conservation. His research over the years has focused on the dimen-
sional behavior of panel paintings and on the packing of works of art for transit. Richard has
served as cochairman of the International Council of Museums (ICOM) Working Group for the
Care of Works of Art in Transit, and as cochairman of the ICOM Working Group for Preventive
Conservation.
Andrea Rothe has been conservator of paintings at the J. Paul Getty Museum since 1981. Born of
German parents in Bolzano, Italy, he grew up in France and Spain during World War II. After the
war he immigrated to the United States with his parents and attended school in North Carolina,
New York, Florida, and Connecticut. After having been accepted into New York Universitys his-
tory of art program, he left with his parents for a trip to Europe. There an introduction by
George L. Stout enabled him to begin an internship at the Uzi Gallery in Florence that changed
the course of his career. He worked rst with a gilder, Rafaello Bracci, and then with the restorers
Augusto Vermehren, Gaetano Lo Vullo, Leonetto Tintori, and Alo Del Serra. He subsequently
did internships with Hermann Lohe at the Bavarian State Galleries in Munich and with Josef
Hajsinek and Franz Sochor at the Kunsthistoriches Museum in Vienna. After this period of train-
ing, he started working on contract for the Italian state in Florence, Naples, Urbino, Arezzo, and
Siena. During this time he also became an assistant to Oskar Kokoschka at his summer academy,
called the School of Vision, in Salzburg.
Ulrich Schiessl received a Ph.D. in art history from Ludwig-Maximilians University in Munich in
1978, and an M.A. in the conservation and restoration of easel paintings and polychrome sculp-
ture from the Academy of Fine Arts in Stuttgart in 1981. Since 1983 he has been a professor in the
conservation department at the Academy of Fine Arts in Dresden. He is a member of ICOM and
an honorary member of the Swiss Association for Conservation and Restoration.
Arno P. Schniewind received B.S., M.W.T., and Ph.D. degrees in wood technology from the
University of Michigan, Ann Arbor. He joined the Forest Products Laboratory, University of
California, Berkeley, in 1956. Initially his specialty was the mechanical behavior of wood and
wood-based materials, and he taught undergraduate and graduate courses and did research in this
area. In 1982 he became interested in the application of wood science to the conservation of
wooden artifacts, and he has since published a number of research papers on that topic. Since his
early retirement in 1991, he has been professor emeritus and continues to be active.
Charles S. Tumosa has a Ph.D. in chemistry from Virginia Polytechnic Institute and State
University. He has twenty-three years of experience running analytical laboratories and has spent
his career examining and determining the character of materials. At present he is a senior research
chemist at the Smithsonian Institution, where he works on the chemical and mechanical proper-
ties of cultural objects.
Luca Uzielli is a professor of wood technology and forest operations at the University of
Florence, Florence, Italy. He specializes in the evaluation, restoration, and conservation of
wooden artifacts, including supports of panel paintings, sculptures, and load-bearing timber
structures of artistic and historical signicance.
560
Zahira Vliz trained as a conservator of paintings at the Intermuseum Conservation Laboratory,
receiving an M.A. from Oberlin College in 1978. She began working as a freelance conservator in
Spain, collaborating extensively with the World Monuments Fund and the Royal Foundation of
Toledo, and working in the conservation department of the Prado Museum in Madrid. She has
taught at the Courtauld Institute, London, and at University College, London. Since 1990 she has
been working privately in London and Spain, lecturing and writing on technical aspects of
sixteenth- and seventeenth-century Spanish painting.
Jrgen Wadum received a bachelors degree in art history from the University of Copenhagen in
1980. He graduated with his B.Sc. (1982) and M.Sc. (1987) in conservation from the School of
Conservation, Copenhagen. He has worked as a freelance paintings conservator since 1983 and
worked at Rosenborg Palace, Copenhagen, in 198788. In 1989 he was employed as a lecturer at
the School of Conservation and at the University of Copenhagen. From 1990 to the present, he
has been chief conservator of paintings at the Royal Picture Gallery Mauritshuis, The Hague. He
is also the coordinator of the International Council of Museums Committee for Conservation
Working Group on Scientic Study of Paintings: Methods and Techniques.
Philip Walker is a practical worker in wood with a special interest in the historical and contem-
porary use of tools and techniques of various trades and cultures. He has written and lectured
widely on these subjects. In 1987 he was elected a fellow of the Society of Antiquaries of London.
He is president of the Tool and Trades History Society, founded in 1983, which has members in
eighteen countries worldwide.
Donald C. Williams has been a furniture restorer and conservator since 1972; his particular
interests are in coating and adhesive materials. He received a B.A. in the technology of artistic
and historic objects from the University of Delaware and joined the Conservation Analytical
Laboratory (CAL) of the Smithsonian Institution in 1984 as furniture conservator. He later
became senior furniture conservator. He is currently coordinator of education and training
programs at CAL.
Antoine M. Wilmering received his training in furniture conservation under the aegis of the
State Training Programme for Conservators in the Netherlands. He was awarded internships in
the Historical Museum of Amsterdam and the Victoria and Albert Museum in London and
received his certicate in Furniture Conservation from the Ministry of Culture in 1983. He was
furniture conservator at Rijksmuseum Paleis Het Loo in the Netherlands before being hired in
1987 by the Department of Objects Conservation at the Metropolitan Museum of Art, New
York, to direct the conservation treatment of the Gubbio studiolo. He is conservator at the
Metropolitan Museum of Art and is responsible for overseeing the work of the furniture conser-
vation sta in the Sherman Fairchild Center for Objects Conservation.
561
Front and back covers: Courtesy of The J. Paul Getty Museum, Los Angeles; photos by
Louis Meluso.
Part One
Opening page: Courtesy of R. Bruce Hoadley.
Hoadley Chemical and Physical Properties of Wood
Figures 113: Courtesy of the author.
Hoadley Identication of Wood in Painting Panels
Figures 129: Courtesy of the author.
Klein Dendrochronological Analyses of Panel Paintings
Figures 115: Courtesy of the author.
Blanchette Guide to Wood Deterioration
Figures 18: Courtesy of the author.
Hanlon and Daniel Modied Atmosphere Treatments
Figures 18: Courtesy of the authors.
Schniewind Consolidation of Wooden Panels
Figures 1, 2: Courtesy of the author.
Part Two
Opening page: J. A. Roubo, Lart du menuisier (Paris: Acadmie Royale des Sciences, 1769).
Uzielli Historical Overview
Figures 1, 2, 6, 9, 1315, 1719, 21, 23, 27: Courtesy of Opicio delle Pietre Dure, Florence.
Figures 3, 4, 1011, 20, 22, 24, 25, 28: Courtesy of the author; drawings by Camilla Burresi
and Daniele Bino. Figure 5: Courtesy of the author. Figures 7, 8, 12, 16: Courtesy of the Uzi
Gallery, Florence. Figure 26: Courtesy of the author; drawing by Camilla Burresi and Daniele
Bino, after drawing in Bomford et al. 1989.
Vliz Wooden Panels and Their Preparation for Painting in Spain
Figures 14, 78: Courtesy of the author. Figures 5, 6: Courtesy of Derek Johns Ltd., London.
Wadum Historical Overview of Panel-Making Techniques
Figure 1: Courtesy of the Academy of Fine Arts, Warsaw. Figures 2, 3, 15, 16, 19ab: Courtesy of
The Royal Danish Collections, Copenhagen. Figure 4: Courtesy of Sothebys. Figures 5ad, 6,
7ah, 14ad: Courtesy of the author. Figures 8, 20: Courtesy of Statens Museum for Kunst,
Copenhagen, Department of Conservation. Figures 9, 10, 17, 23, 25: Courtesy of The Royal
Picture Gallery, The Mauritshuis, Conservation Department, The Hague. Figure 11: Courtesy of
the Nederlands Kunsthistorisch Jaarboek 1978. Figure 12ab: Courtesy of The National Gallery,
563
Illustration Credits
London. Figure 13: Courtesy of the Muse des Beaux-Arts de Rennes. Figure 18: Courtesy of the
Muse des Beaux-Arts de Dijon. Figure 21: Courtesy of Christies London. Figure 22: Courtesy of
Christies Amsterdam B.V. Figure 24: Courtesy of Instituut Collectie Nederland, Netherlands
Institute for Cultural Heritage.
Walker History of Relevant Woodworking Tools and Techniques
Figure 1: Courtesy of the Bibliothque Municipale de Dijon. Figure 2: Courtesy of the
Cathdrale Notre Dame de Chartres. Figure 3: With permission of The Casa Editrice
Francescana, Assisi. Figures 46, 10, 12ac: Courtesy of the author. Figure 7ab: Courtesy of
the Early American Industries Association. Figure 15: Courtesy of Dr. G. Heine and the Tool
and Trades History Society.
Part Three
Opening page: Courtesy of The J. Paul Getty Museum, Los Angeles; photo by Louis Meluso.
Rothe Critical History
Figures 1, 2, 5, 7, 8: Courtesy of the Ministero per i Beni Culturali e Ambientali, Soprintendenza
per i Beni Ambientali Architettonici, Artistici e Storici di Arezzo. Figures 3, 11: Courtesy of the
Ministero per i Beni Culturali e Ambientali, Soprintendenza per i Beni Artistici e Storici delle
Marche, Urbino. Figure 4: Courtesy of the National Gallery, London. Figure 6: Courtesy of the
Ministero per i Beni Culturali e Ambientali, Soprintendenza per i Beni Artistici e Storici di
Bologna. Figures 9, 10: Courtesy of The J. Paul Getty Museum, Los Angeles, California. Figure 12:
Courtesy of the author. Figures 1320: Courtesy of the Ministero per i Beni Culturali e Ambientali,
Istituto Centrale per il Restauro, Rome. Figures 21, 22: Courtesy of the Sarah Campbell Blaer
Foundation, Houston, Texas.
Schiessl History of Structural Panel Painting Conservation
Figures 13: Photos by W. Rabich; courtesy of the Landesamt fr Denkmalpege Sachsen,
Dresden. Figures 4, 17, 18: Courtesy of the Landesamt fr Denkmalpege Sachsen, Dresden.
Figures 5, 19: Courtesy of the Schweizerisches Institut fr Kunstwissenschaft, Zurich. Figures 6,
7, 12: Courtesy of the Bayerisches Nationalmuseum, Munich. Figures 8, 9: Courtesy of the
Kunstmuseum Berne. Figure 10: Courtesy of the Hochschule fr Bildende Knste, Dresden.
Figures 11, 1316, 20, 21: Courtesy of the Staatliche Kunsthalle Karlsruhe.
McClure History of Structural Conservation of Panel Paintings in Great Britain
Figures 13, 611: Courtesy of the Hamilton Kerr Institute. Figures 4, 5: Courtesy of Kingston
Lacy, The Bankes Collection (The National Trust). Figures 1215: Courtesy of Royal Collection
Enterprises.
Bret, Jaunard, and Mandron Conservation-Restoration of Wooden Painting Supports
Figure 1: Photo Routhier; courtesy of the Service de Restauration des Muses de France.
Figure 2: Photo by J. Requil and D. Jaunard; courtesy of the Service de Restauration des
Muses de France. Figures 3, 4: Photos by D. Jaunard; courtesy of the Service de Restauration
des Muses de France. Figures 5, 6: Photos by G. Dufresne; courtesy of the Service de
Restauration des Muses de France. Figures 7, 8: Photos by J. Requil; courtesy of the Service
de Restauration des Muses de France.
Bergeon et al. Two Hundred Years of History In France
Figures 1, 2, 1214, 16: Courtesy of the Louvre Museum. Figures 311, 15, 17: Courtesy of the
Muse du Petit-Palais, Avignon.
Horns Richard Buck
Figures 114: Courtesy of the Straus Center for Conservation, Harvard University Art Museums.
Part Four
Opening page: Courtesy of The J. Paul Getty Museum, Los Angeles; photo by Louis Meluso.
564
Rothe and Marussich Florentine Structural Stabilization Techniques
Figure 1: Courtesy of the Ministero per i Beni Culturali e Ambientali, Soprintendenza per i Beni
Ambientali Architettonici Artistici e Storici di Arezzo. Figures 2, 58: Courtesy of the authors.
Figures 3, 4, 1012: Courtesy of The J. Paul Getty Museum, Los Angeles. Figure 9: Courtesy of
the Staatliche Museen zu Berlin, Preussischer Kulturbesitz, Gemldegalerie.
Castelli Restoration of Panel Painting Supports
Figures 128: Courtesy of the Ministero per i Beni Culturali e Ambientali, Opicio delle Pietre
Dure di Firenze, Florence.
Bisacca A Nativity by Francesco di Giorgio Martini
Figures 1, 4, 5, 15, 2533: The Metropolitan Museum of Art, New York. Figures 2, 3, 713,
1624: 1997 Board of Trustees, National Gallery of Art, Washington, D.C. Figures 6ac, 14b:
The Metropolitan Museum of Art. Drawings by Daniel Kershaw. Figure 14a: Courtesy of the
Ministero per i Beni Culturali e Ambientali, Soprintendenza per i Beni Artistici e Storici di Siena.
Lebas Cradling of a Relief Attributed to Martin Schaner
Figures 15: Courtesy of the Museum fr Kunst und Gewerbe, Hamburg.
Glatigny Backings of Painted Panels
Figures 1, 6: Courtesy of the Muse Athois, Ath, Belgium. Figures 25: Courtesy of the author.
Bobak Flexible Unattached Auxiliary Support
Figures 114: Courtesy of the author.
Marchant Development of a Flexible Attached Auxiliary Support
Figures 14, 6: Courtesy of the author. Figures 5, 718: KeirincxSaveryOld Franks, Orpheus
Attacked by the Thracian Women [Death of Orpheus], from a private collection.
Reeve Structural Conservation of Panel Paintings at the National Gallery, London
Figures 113: Courtesy of the Trustees of the National Gallery, London.
Brewer Some Rejoining Methods
Figures 13, 59: Courtesy of the author. Figure 4ab: Private collection, Scotland.
McClure Framing of Wooden Panels
Figures 1, 3, 68, 1016: Courtesy of the Hamilton Kerr Institute, Fitzwilliam Museum, University
of Cambridge. Figures 4, 5, 9: Courtesy of Kingston Lacy, The Bankes Collection (The National
Trust). Figure 2: The Chapel of Our Lady and St. Margaret, Oxburgh Hall (The National Trust).
Brewer Practical Aspects
Figure 1, 4c, 13: Courtesy of the author. Figures 2, 3, 7: Courtesy of the Hamilton Kerr Insitute,
Cambridge. Figures 4ab, 14, 16ab, 17: The Property of the Marquess of Northhampton. Figure
5ab: Courtesy of the Tate Gallery, London. Figure 6ac: By kind permission of the Provost and
Scholars of Kings College, Cambridge. Figure 8ae: Reproduction by pemission of the Syndics of
the Fitzwilliam Museum, Cambridge. (Figures 8b, 8d, 8e, photographs by Christopher Hurst.)
Figures 9ac, 10ab, 11ab, 12: Courtesy of the Warden and Fellows of All Souls College, Oxford.
Figure 13: Courtesy of the J. Paul Getty Museum, Los Angeles. Figure 15: Courtesy of the
Wallraf-Richartz Museum, Cologne.
Wilmering A Renaissance Studiolo from the Ducal Palace in Gubbio
Figures 13, 519: The Metropolitan Museum of Art, Rogers Fund, 1939 (39.153). Figure 4:
The Metropolitan Museum of Art. Drawing by Daniel Kershaw.
Wadum Microclimate Boxes for Panel Paintings
Figures 13: Courtesy of the author.
Richard, Mecklenburg, and Tumosa Technical Considerations for the Transport of Panel Paintings
Figures 126: Courtesy of the authors.
565
P R O C E E D I N G S
P
R
O
C
E
E
D
I
N
G
S
The Ge t t y Cons e r va t i on I ns t i t ut e
The Getty Conservation Institute
Los Angeles
T
h
e
S
t
r
u
c
t
u
r
a
l
C
o
n
s
e
r
v
a
t
i
o
n
o
f
P
a
n
e
l
P
a
i
n
t
i
n
g
s
Proceedings of
a Symposium
at the J. Paul
Getty Museum,
April 1995
The Structural
Conservation of
Panel Paintings