1 INTRODUCTION

The structure of a panel painting typically includes layers of a number of different materials over the wooden support. The support of a traditional panel painting was sometimes first prepared with a layer of fabric and virtually always prepared with a ground layer. The design layers usually include: a preparatory design that was drawn and/or incised on the ground, possibly a metal leaf layer, the paint layer, the varnish, and any retouchings. Later artists also have used wood products such as hardboard for supports, with surface layers that can include a ground, fabric prepared with a ground, and paper.

Unrestrained wood will expand and contract. If the wood expands while adsorbing water but is restrained because of the panel construction, compression of the wood structure can occur; the cell walls collapse and do not regain their shape upon desiccation. When a restrained panel loses water during a desiccation period, local tensile stresses can develop that exceed the strength of the wood. Both processes can lead to warping and splitting. When a panel painting is restrained by a cradle, a pattern of damage (often called “washboarding”) can result; the panel is forced to distort into a series of independent warps between splits induced by the fixed cradle members.

The stresses due to dimensional changes can also result in delaminations and cleavage within the various layers of the painting, which cannot always follow the movement of the support. The ground can separate from the support; paint layers can separate from the ground or from other paint layers. Even the support itself can delaminate; composite supports may delaminate between a surface layer and the lower support (for example, illustration board over hardboard), or delaminations can occur within the support itself (for example, within hardboard, which is a human-made, fibrous material).

Internal flaws cannot always be detected either visually or by traditional testing techniques such as tapping or radiography. Infrared thermography reveals flaws, depending upon the material properties of the sample (diffusiv-ity, conductivity) and the location and size of a flaw (Murray 1990). Radiography techniques (especially x-ray radiography, but also xeroradiography) have been widely used to determine the condition of paintings as well as their construction and any previous repairs.

The air-coupled ultrasound imaging approach used here differs from past ultrasonic work on art objects, which used contact transducers with gels or pressure to examine stone, metal, and waterlogged wood and took measurements at specific points rather than over complete areas (Miura 1976; Asmus and Pomeroy 1978; Canella et al. 1983; Chiesura et al. 1983; Pilecki and Levi 1983; Rossi-Manaresi and Tucci 1983; Berra et al. 1988). The mechanical scanner used here allows entire paintings to be examined with ultrasound and with no contact; air is used as a couplant and therefore no residues are left. The two-dimensional images produced permit comparisons to be made between local and adjacent points (Murray et al. 1991; Murray 1993; Murray et al. 1994).

In this study, a through-transmission configuration is employed in which a transducer is positioned on either side of the sample; both sides of the sample, therefore, need to be accessible. One of the transducers transmits a low-amplitude stress wave and the other receives it. The received signal can then be interpreted to deduce the properties and condition of materials. In an unflawed sample, the signal has a continuous path along which to travel; how-ever, when it encounters air pockets, the signal can be reduced in strength, slowed down, or completely lost. This change in signal character is used as a discriminant of internal anomalies such as splits, checks, delaminations, cleavage, or voids.

Signal characteristics include the amplitude, time of flight, and phase. The amplitude, the vertical height of the envelope of the received signal, is reduced to zero in delaminated areas. The time of flight is the amount of time taken for the signal to transverse the material, which is an indication of its properties and thickness. The phase, also a measurable property of the received signal, is, in this case, dependent upon the velocity of the wave through the material and can be used to detect splits (Murray 1993). The air-coupled ultrasound method is now possible because of newly available noncontact transducers and electronics that produce stronger signals.

The ability of any signal to traverse boundaries (for example, between air and the material) depends upon the acoustic impedances (the material's density multiplied by the material's wave speed). Materials that can be investigated with air-coupled ultrasound must have a relatively low acoustic impedance so that the difference between air (which has a low value) and the material is not too great. Examples of materials that can be examined include ligneous materials, foams, fiber-reinforced composites, rubber, paper, and nonmetallic composites. A similar air-coupled ultrasound technique has located delaminations within hardboard, insulation board, and particle board (Birks 1972). The ultrasound signal of the system discussed in this paper traversed different thicknesses of wooden panels, the thickest being 1.6 cm for oak, 3.5 cm for poplar, and 0.6 cm for hardboard.

Radiography reveals voids and splits oriented parallel to the x-ray beam and perpendicular to the surface of the painting, because there is a large discontinuity in the material and therefore a change in density. However, when a flaw is oriented parallel to the surface of the painting, as in the case of blind cleavage, the change in density is generally too small for radiography to detect. In this case ultrasound is useful because the sound wave is stopped and the signal amplitude is registered as zero. When the flaw is at an angle, both techniques can be used to advantage.