2 2. METHODOLOGY

The technical investigation involved three stages: stereomicroscopy, nondestructive analytical techniques including xeroradiography and energy dispersive x-ray fluorescence (ED-XRF), and finally scanning electron microscopy with both wavelength dispersive spectrometry (SEM-WDS) and energy dispersive spectrometry (SEM-EDS).

Visual examination was undertaken with a binocular optical microscope (10–40x). The cross sections of Ptolemaic faience at broken edges were examined, and a layered structure—body, an underglaze, and a glaze—was defined.

Xeroradiography, a nondestructive analytical technique, was used to investigate the construction method of Ptolemaic faience.2 Xeroradiography is capable of detecting minor differences in the radiodensity of solid objects. It has particular application in examining ceramics, as it produces images of greater contrast than conventional x-ray radiography. Joints, seams, air bubbles, inclusions, and the orientation of the ceramic body can be easily identified, all of which are not usually visible in conventional x-ray radiography.

Another nondestructive analytical technique, qualitative ED-XRF analysis, was chosen for initial determination of the elemental composition of the surface of the glaze as well as that of the underglaze.3 The goal was to group the 21 pieces by similarity in the glaze and the underglaze constituents in order to select a group of representative pieces for sampling for scanning electron microscopy. Since ED-XRF provides only surface analyses, it is possible that the x-rays may not penetrate through the glaze layer to reach the underglaze. Therefore, the selected sites for the underglaze analyses included areas where the glaze was worn or chipped away, exposing the underglaze. The purpose of ED-XRF analyses of the underglaze was to check whether there were additional elements, not found in the glaze.

Based on the ED-XRF results and stylistic diagnoses, five fragments were selected (acc. nos. 48.376, 48.377, 48.379, 48.380, and 48.386) for SEM analyses to represent the group of 21 pieces (fig. 3). The decorative motifs of these five pieces are commonly seen in Ptolemaic faience. They can be linked to low-relief and bicolored Ptolemaic faience in other collections for future comparison. The range of glaze colors, in terms of variation in shades of blue and green, was also a consideration in choosing the samples.

Fig. 3.

Five fragments from which samples were taken for SEM analyses, Walters Art Gallery, Baltimore, acc. nos. 48.376, 48.377, 48.379, 48.380, 48.386

The primary analytical method employed to characterize major and minor constituents in the glaze, the underglaze, and the body was scanning electron microscopy. Polished sections of the samples were made to include the bodies, the underglaze layers, and the glaze layers.4 Some samples included only the exterior surface, and some included both interior and exterior surfaces. The description of each sample is reported in table 2. Scanning electron microscopy-wavelength dispersive spectrometry (SEM-WDS) was undertaken for quantitative analyses of the glaze and the underglaze. WDS was chosen for its accuracy and precision in analyses. Its high resolution minimizes the potential for any overlap of analytical x-ray lines in the study of ancient glassy materials including glazes (Verita et al. 1994). Three spots were selected at random for repetition of the WDS analysis in each of the glaze and the underglaze layers. The averages and standard deviations (1-sigma) of the variability of the three analyses are reported in table 3. All WDS results were reported in the actual range of oxide-weight percentages and have not been normalized. Scanning electron microscopy-energy dispersive spectrometry (SEM-EDS) was undertaken for semiquantitative analysis of the overall body composition. EDS allows the analysis of a large area, which is important in heterogeneous samples (fig. 3). Only one site per sample was analyzed. In all cases, a maximum body area of a sample was analyzed, up to several square millimeters. All EDS results were normalized and are reported in the actual range of oxide-weight percentages in table 4. They were obtained from the average value of the area analyzed. The operating conditions used for EDS and WDS analyses are described in the appendix.