E. TRINKAUS 



(Eschman, 1992) of SLICE (Nagurka & Hayes, 1970). All sections 

 were digitized twice and the results averaged. 



From this, six primary measurements were computed. These 

 included total subperiosteal (TA) and cortical (CA) areas, from 

 which medullary area (MA) can be computed, as well as the second 

 moments of area relative to the antero-posterior (I ) and medio- 

 lateral (I ) axes, the maximum second moment of area (I ), and the 

 perpendicular to I (I ). The polar moment of area (J, or I ), a 

 measure of torsional rigidity and overall strength, is the sum of any 

 two perpendicular second moments of area (usually I + I , but 

 also equal to I x + I ) 



For a few of the Mesolithic comparative specimens, subperio- 

 steal contour molds were unavailable. For these, the 

 cross-sectional parameters were computed using standard ellipse 

 formulae (Runestad et al., 1993) from the subperiosteal diameters 

 and cortical thicknesses. Given the antero-posterior and medio- 

 lateral orientations of the radiographs, the resultant cross-sectional 

 measures include only cross-sectional areas and antero-posterior 

 (I ) and medio-lateral (I ) second moments of area, plus the polar 

 moment of area computed as the sum of I x and I . For these, the 

 resultant computed values were corrected for parallax and non- 

 ellipse shapes of the cross-sections using least squares regressions 

 between the radiographically determined measurements and the 

 cross-sectional values obtained from digitizing the same sections 

 of the other Mesolithic femora or tibiae. 



To assess proportions in the Gough*s Cave 1 diaphyses using 

 cross-sectional parameters, three shape indices were computed, 

 percent cortical area (%CA: (CA/TA) x 100), I/I (as a ratio) and 

 I /I . (also as a ratio). The last two assess diaphyseal shape at the 

 cross section locations, the former with respect to the anatomical 



axes and the latter with respect to the axis of maximum bending 

 rigidity. The second is especially appropriate in the proximal femo- 

 ral diaphysis and along the tibial diaphysis, given varying degrees of 

 torsion in the proximal epiphyses of these bones. 



To assess robusticity, and hence to scale the cross sectional 

 parameters to appropriate body size and beam characteristics (Ruff 

 etal., 1993), cortical areas (as a reflection of axial loading levels) and 

 polar moments of area (as a measure of resistance to bending and 

 torsional loads) should be plotted against appropriate powers of long 

 bone lengths adjusted for variance in body laterality and crural 

 indices. Cortical areas should scale to body mass, which is propor- 

 tional to femoral length cubed (Ruffe? a/.. 1993). Polar moments of 

 area should scale to body mass times beam length, all raised to the 

 four-thirds power (Ruffe? al., 1993). In other words, for the femur J 

 °c (FL 3 xFL) 4/, = FL 16/3 andforthe tibia J°c (FL 5 xTL) 4/, = FL 4 xTL 4 '\ 



However, given the apparently similar degrees of body laterality 

 and crural indices across these European terminal Upper Paleolithic 

 and Mesolithic samples, as is expected by theoretical considerations 

 (Ruff, 1991)and supported by current data(Holliday, 1995;Holliday 

 & Churchill. 2003). it is appropriate to simply scale logged cortical 

 areas and logged polar moments of area against logged bone length. 

 Since this approach avoids determining the actual allometric scaling 

 coefficient for each of these bones, it is employed here. 



In addition, even though comparative data are not available, 

 metatarsal midshaft cross-sectional geometric measures are pro- 

 vided (Table 20). They were computed from radiographically 

 determined subperiosteal diameters and cortical thicknesses using 

 ellipse formulae (Runestad et al., 1993) after the radiographic meas- 

 urements were corrected for parallax using the osteometrically 

 determined diaphyseal diameters. 



Fig. 1 Ventral (left) and dorsal (right) views of the sacrum; x 0.75. 



