This approach seemed to work well, though the original wood-ground interface was lost.
Epoxy fillers, which would not swell the wood so much, were considered, but penetration and
flow are hard to control. Also, most cured epoxies are mechanically intractable to the move-
ment ofsurrounding wood. They are also too efficient as moisture barriers, and adhesion of
other glues is limited (Skeist 1977:chap. 26). Rather than the epoxy, something like a strong,
flexible, 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
fibrous, lower-density wood of the Palmezzano, which swelled less from the same glue and
did not transfer the swelling detrimentally. There was no apparent effect on the strongly
adhered gesso ground ofthe 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 fitted carefully to a panel’s surface, are usually glued to the
high spots only. Aside from the inherent restrictions on transverse wood movement from any
glued restraint, the intermittentattachment also helps to localize and concentrate tensions
due to restraint of differential 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 30–90% at 25 °C(Building Research Establishment 1975:6).
40 See Gordon’s absorbing discussion of “critical Griffith crack length” (Gordon 1978:chap. 5,
esp. 98–105).
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 effect 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 Pietà de
Villeneuve-lès-Avignon(1454–56) by Enguerrand Quarton (Louvre INV RF1569), also painted on
a walnut support (Bergeon 1990:35–38).
46 Animal glue, for example, can form a substantial restraint.
47 Warp from this cause is modified by movement deriving from the cuts of the planks and by
any restraint caused by applied layers and by joints.
48 Some effects offlattening 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 werediscussed theoretically, the practical application and consequences were
not conclusive. Regarding “slippage” and flattening, Buck states that his conclusions about slip-474 Brewer