CT Scanning: The specimen was scanned at the High-Resolution X-ray CT Facility of the University of Texas at Austin on July 29, 1999. The slab was mounted upright with the longest axis vertical and CT slices were taken through the bone-bearing portion of the slab perpendicular to its face. Scanning was performed using a Pantak high-energy X-ray source at an energy level of 420 kV and 4.7 mA, and a spot size of 1.8 mm. The X-ray beam was pre-filtered using a 1.58-mm brass plate. Attenuated X-ray energies were detected by a 512-channel solid-state P250D linear array of CdWO4 scintillating crystals that records across a 12-bit dynamic range, and the 1.0 slice thickness was set using tungsten collimators. The scan geometry was 3rd-generation 190% offset mode13, with each slice consisting of 2000 84-ms views spanning a 360° rotation. Reconstruction parameters were optimized to utilize the full 12-bit grayscale range available in the output images. The reconstruction field of view for each image was 270 mm, and slices were collected at 0.9 mm intervals (i.e., 10% slice overlap).

Image Processing: The CT slice images were collected in 16 bits at a size of 1024 by 1024 pixels and later cropped using Adobe Photoshop software to eliminate uninformative pixels around the periphery of the skeleton and to reduce image file size. The imagery was exported as both 16-bit and 8-bit TIF files and archived on CD-ROM. To facilitate inspection of the entire volume, the datasets were synthetically re-sliced in horizontal planes using NIH Image software. Digital animations through the consecutive serial section stacks were generated with Quicktime software, and a 3-D volumetric model of each slab was generated using Voxblast software. Reduced versions of these datasets and visualizations were prepared for Web delivery (see supplementary information). The specimen was also photographed under high magnification using ultraviolet light.