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 3 162.5 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 fixing of the panel in its frame would have inevitably produced rup-
ture ofthe wood or of the glued joins.^9 Stout, in describing a formula to
calculate the force required to constrain and flatten 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 1955:158–59). In practice, however,
evaluation of the force required for the panel to deflect the frame fixings
seems only recently to have been assessed. In 1991 Mecklenburg and
Tumosa produced computer models of cracked and uncracked oak panels
rigidly fixed into their frames and assessed their resistance to splitting. An
uncracked oak panel measuring 76 3102 3 1.27 cm thick could be split by
fluctuations of RH between 70% and 10%. When a cracked panel ofsimi-
lar dimensions is subjected to strain, much less force is required to extend
the cracks (Mecklenburg and Tumosa 1991:187ff.). 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 byflexible spring fixings, commonly used to secure panel paintings,
or by malleable brass fixings, 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^23 are
commonly used).^10 However, a simple experiment demonstrates that the
force required to deflect a particular fixing is much greater than might be
expected. Two commonly used sprung-steel fixings and four brass fixings
of different dimensions were screwed to a length of wood. Holes (or in
the case of one spring fixing, a hook) were provided to attach a spring
balance. The force required to raise a fixing by 1 cm was observed. The
length ofeach fixing, which affects the moment of the force generated,
was not assessed. The spring fixings required a loading of 0.8–1.4 kg to
deflect 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 fixing, a result that reflected
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^23 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 5mm. Two identical blocks at 10 cmcenters 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 5mm. The force exerted by the metal fixing 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
T F W P 439