and 3 are slightly positively skewed (Table 12). However, the appreciable 

 negative skewing that characterized Vallois's combined-sex sample of 260 

 capacities is not a feature of my own combined-sex pooled sample of 402 

 capacities. The latter distribution has maximum and minimum values 

 almost exactly 3 S.D.s above and below the mean (as is to be expected 

 in a normally distributed variate when ji = 442). 



The gorilla samples are strongly positively skewed in males and 

 slightly in females. The differences between the deviations of the extreme 

 values are +1.66 S.D.s for the latest series of total pooled males, +0.53 

 S.D.s for total pooled females, and +1.25 S.D.s for a combined-sex pooled 

 sample. 



Similarly, the modern human sample is definitely positively skewed, 

 2 outside estimates giving differences between the extreme deviations of 

 + 1.05 and +1.25 S.D.s. 



Thus, 4 of the hominoid samples cited, namely gibbon, siamang, 

 chimpanzee, and orangutan, have virtually symmetrical distributions; and 

 2, gorilla and modern man, have positively skewed distributions. 



The meaning of the variable distribution of cranial capacities. 

 The different kinds and degrees of skewing that the figures in Table 12 

 have demonstrated pose an interesting problem. 



In short, the right-skewed distributions of human and gorilla values 

 indicate that variants with large capacity are more common than variants 

 with small capacity in the samples representing the populations. Accord- 

 ing to Simpson et al., such moderately right-skewed distributions are 

 common in zoology and "although an adequate census has not been 

 made, they appear to be more common than are left-skewed distributions" 

 (Simpson, Roe, and Lewontin 1960, p. 144). Contrariwise, the left-skewed 

 distributions of capacities for some orangutan samples, as well as for 

 some other hominoid series, show that small variants are more common 

 in these samples than large variants. In most gibbon, siamang, chim- 

 panzee, and orang samples the small and large variants occur in approxi- 

 mately equal frequencies. 



Do these differences of distribution within a series of big hominoid 

 samples reflect real population differences? At present. I am inclined to 

 assume the null hypothesis and to accept that the different kinds and 

 degrees of skewing result mainly from peculiarities of sampling rather 

 than from population differences in the pattern of distribution of the 



57 K 



