Fig 1.
Front and back of the Rijksmuseum’s panel painting Cadmus Sowing Dragon’s Teeth (http://hdl.handle.net/10934/RM0001.COLLECT.5320).
The strips of oak that surround the panel can be observed on the photo of the reverse.
Fig 2.
Possible orientations to scan the panel.
Horizontal tiling (left): this setup requires fewer tiled scans, but the rotation axis lies outside the field of view of the detector for all tiles except the central one, resulting in parts of the painting to rotate in and out of the field-of-view. Vertical tiling (right): this setup requires more tiles, but the section of the painting we want to image stays within the detector field-of-view for all tiles during the rotation.
Fig 3.
Scanning the Cadmus painting at the FleX-ray Lab.
(A) Panel on the Ethafoam® mount placed on the rotation stage at approximately 90 degrees rotation. The vertical tiled scans are achieved by moving the source (on the left) and detector (on the right) up or down by the same distance, so that they stay directly opposite to each other; (B) X-ray image of Tile 2 at 0 degrees. The foam can be seen at the bottom, whereas the slightly darker area above represents the oak strip; (C) X-ray image of Tile 2 at 90 degrees. The X-rays must travel through approximately 44 times more wood in this direction. The nails that fix the oak strip to the panel can be seen.
Fig 4.
(A) Reconstructed CT image of the full cross-section of the panel, providing an overall impression of the configuration of the panel and the transverse section of the oak back board, which has been processed radially from the parental tree; (B) reconstructed CT image of the Tile 2. Tile 2 corresponds to the left edge of the panel in (A), and the reconstructed image (with 37.3 micron resolution) shows: the oak strip on the left; the paint layer on the upper part, with white areas caused by highly absorbing metal-containing pigments in the paint; the original wooden painting support, which appears to be of a diffuse-porous species; and a second (oak) board glued on the back of the original board. The oak board was initially the target of this investigation as it was assumed to be the original support. Tree rings in the oak board are evident in this image (growth direction towards the left in both images).
Fig 5.
Species identification of the front (original) support.
(A) direct observation of wood anatomical features with the Hirox RH-2000 digital microscope; (B) removal of a micro sample of wood from the original support in the top-right corner for wood identification. Key anatomical features characteristic for Swietenia sp. [52–54]: (C) Transverse section, 50x: large vessels 100–270 μm, diffuse-porous, mostly solitary, occasionally in pairs; gums in vessels (1); parenchyma apotracheal and paratracheal; terminal parenchyma are not visible due to small sample size. (D) Radial section, 200x: septate fibres (2) present; gums in parenchyma. (E) Radial section, 200x: Simple perforation plates; numerous vessel ray pits (3) alternate, simple, very small 1.9 to 3.5μm. (F) Radial section, 500x: cross-field pits small and bordered (4), prismatic crystals in marginal (procumbent) ray cells (5). (G) Tangential section, 200x: Rays 1–4 cells wide (6), 9–20 cells high, slightly heterogeneous, slightly storied. (F) Tangential section, 500x: prismatic crystals in marginal (procumbent) ray cells (7).
Fig 6.
Results of the dendrochronological research.
(A) Cross-match between the tree-ring series obtained from each individual tile. (B) Absolute dating of the mean curve obtained from averaging the measurements of the individual tiles (Cadmus) crossdated with the Baltic2 chronology. TBP, Student’s t-value as implemented by [60] for tree-ring studies; THO, Student’s t-value as implemented by [43]; r, correlation coefficient; Gl, percentage parallel variation between the overlapping portion of the compared tree-ring series [61], accompanied by its signification level; Ol, overlap (see S3 Text for details about the statistical tests used).