GOUGH'S CAVE 1 : ASSESSMENT OF BODY SIZE AND SHAPE 



41 



lines. As Ruff has found, there is good separation of the Europeans 

 from the Sub-Saharan Africans throughout most of the size range. 

 Informatively, Gough's Cave 1 falls squarely on the European re- 

 gression line, far above the Sub-Saharan African line. 



Multivariate Assessment of Body Shape 



Any assessment of an individual's body size and proportions is at its 

 base an assessment of that individual's total morphological pattern. 

 While the individual analyses presented above when considered as a 

 whole provide tantalizing clues as to the total morphological pattern 

 of the Cheddar Man, these analyses are likely not as informative as 

 would be a multivariate assessment based on the same morphologi- 

 cal variables. In fact, a multivariate analysis may be expected to 

 resolve some of the conflicting results obtained above. For example, 

 in relative body linearity, relative body breadth and limb/trunk 

 (excepting the tibia) proportions, the Gough's Cave specimen looks 

 essentially like a recent European (albeit occasionally at the more 

 linear end of the European range). In contrast, his tibia/trunk, bra- 

 chial, and especially his crural index are more similar to those of 

 more tropically-adapted groups (e.g., Africans). 



What then, is the total morphological pattern of body size and 

 shape exhibited by Gough's Cave 1? The way to discover this is to 

 investigate overall body proportions in multivariate space, taking the 

 variances and covariances of all the skeletal manifestations of body 

 shape into account. Once this is done, Gough's Cave 1 will either 

 continue to fall among recent Europeans, or he could possibly 

 exhibit a somewhat different, more tropically-adapted pattern. 



The variables to be used in the multivariate analysis and their 

 abbreviations are found in Table 7. Note that these measurements are 

 the same variables used to compute ratios and/or which were plotted 

 in bivariate space. They should therefore provide an accurate reflec- 

 tion of total body shape. The method chosen for body shape extraction 

 is that outlined by Mosimann and colleagues (Mosimann & James, 

 1979; Darroch & Mosimann, 1985; James & McCulloch, 1990). 

 These morphometricians argue that an individual's overall size is 

 best represented by the geometric mean of all the measurements 

 taken on that individual. The geometric mean (or 'log size' as the 

 authors denominate it) can then be used to create scale-free ratios, or 

 'shape' variables, between each of the individual's measurements 

 and his geometric mean. The utility of the shape variables lies not in 

 the 'removal' of size per se, but in the ability of the researcher to 

 determine if there is a relationship between size and shape via 

 correlation analyses. The application of this method to anthropologi- 

 cal data sets is discussed in greater detail elsewhere (e.g., Falsetti et 

 ai, 1993; Jungers et ah, 1995). In this study, since the primary 

 interest is the body shape of Cheddar Man, discussion is limited to 

 the analysis of shape variables. The variance-covariance matrix 

 (VCM) of the shape variables for a combined sample of fossil and 



Table 7 First two principal components of shape variables - fossil and 

 recent humans. 



Eigenvector Coefficient 

 I II 



Femoral A-P head diameter (FHAP) 

 Bi-iliac breadth (BIB) 

 Femoral bicondylar length (FL) 

 Humeral maximum length (HL) 

 Tibial maximum length (TL) 

 Radius maximum length (RL) 

 Skeletal trunk height (STH) 



Eigenvalue 



% total variance 



0.305 



-0.860 



0.591 



0.451 



-0.246 



0.070 



-0.178 



0.037 



-0.404 



0.124 



-0.421 



-0.009 



0.591 



0.187 



0.0094 



0.0032 



58.25 



19.63 





<u 



a 

 a 



u 



PCI Shape (58.3%) 



Fig. 2 Scatter plot of PC2 on PC 1 (shape data). Crosses are recent 

 Europeans, open squares are recent Sub-Saharan Africans, triangles are 

 Early Upper Paleolithic, circles are Late Upper Paleolithic, closed 

 squares are Mesolithic, star is Gough's Cave 1. Lines indicate range of 

 the recent human samples. 



recent humans (n = 225), all of whom preserve the measurements in 

 question, was computed and then subjected to principal components 

 analysis (PCA). 



The eigenvector coefficients and eigenvalues for the first two 

 principal components of the log shape data are found in Table 7. The 

 first principal component (PCI) accounts for 58.3% of the total 

 shape variance, and contrasts limb segment length (particularly the 

 distal segments) with femoral head diameter, bi-iliac breadth and 

 skeletal trunk height. The PC scores along this axis are not signifi- 

 cantly correlated with overall size (i.e., the geometric mean; r 2 = 

 0.008, p = 0. 1 758). PC 1 is readily interpreted as a climatic adaptation 

 component, since it separates heavier, less linear individuals (more 

 cold-adapted) from lighter, more linear individuals (more heat- 

 adapted). The second principal component (PC2) accounts for 19.6% 

 of the shape variance and contrasts bi-iliac breadth and trunk height 

 with femoral head diameter. The scores along this axis are correlated 

 with log size (r 2 = 0.18, p < 0.0001), and this component tends to 

 segregate males (who on average have large femoral heads and 

 relatively narrow pelves) from the small femoral-head possessing 

 and wider hip bearing females (albeit with considerable overlap). 



Component scores for the European early modern fossils (includ- 

 ing Gough's Cave 1 ), as well as the recent Sub-Saharan Africans and 

 recent Europeans are plotted in Figure 2. Note that the separation of 

 the groups is along the first principal axis. This axis contrasts 

 individuals on the left, who possess short distal limbs, wide and 

 relatively long trunks, and large femoral heads from those individu- 

 als on the right, who are characterized by relatively short and narrow 

 trunks, long distal limb segments and smaller femoral heads. There 

 is no separation of the groups (fossil or recent) along the second 

 principal axis. Note that for the first principal component, there is 

 relatively little overlap between the recent Sub-Saharan Africans 

 (represented by open squares ) and the recent Europeans (represented 

 by crosses). Gough's Cave 1 (the star) and his contemporaries, the 

 European Mesolithic specimens (indicated by dark squares) fall 

 clearly among the recent Europeans, as do the LUP specimens 



