GROWTH AND MORPHOLOGICAL CHARACTERS. 71 



variable as they approach the guinea-pig, we find such to be the case 

 for the males in measurements 3, 6, and 10, and for the females in 

 measurements 3, 5, and 8 to 16. (The ^ wild are not considered on 

 account of the small numbers.) Now, if we had by chance chosen 

 only measurements 3 and 10 as the basis for our comparisons, then we 

 would have been led to the conclusion that there was consistent evidence 

 of segregation and a recombination of size factors in both sexes. But 

 had we chosen other measurements we might have arrived at different 

 conclusions. The question naturally arises, are the series of coefficients 

 in any one dimension more significant than those in any other ? Are 

 we justified in selecting particular series which conform to the results 

 presented by other investigators, and thus indicate a recombination 

 of factors? As far as we can tell, we are not ; for at present we know 

 of no reason why special emphasis should be attached to the results 

 obtained in certain measurements in preference to others. 



There is another method of approach by which it is possible to avoid 

 attaching questionable weight to a few dimensions. We may average 

 all the coefficients of variabihty in each of the different classes to see, 

 for example, if the | wild were on the whole more variable than the 

 guinea-pig. Table 69 gives the averages of the different coefficients 

 of variability in the guinea-pig and hybrids, the purpose being to 

 ascertain what the general tendencies of any class might be and to 

 see whether on the whole the hybrids showed a general tendency to 

 greater average variability than the parent guinea-pig. We also wished 

 to see if, on the whole, the I wild were more variable than the | wild 

 and the guinea-pig. However, the male | wild averaged no more 

 variable than the guinea-pig; but the female | wild were more variable. 

 All the different classes of males were of equal average variability 

 except the I wild. All the female classes were statistically of equal 

 average variability except the female guinea-pigs. The males do not 

 show a series indicating that the f wild average most variable and 

 that this variability decreases as we approach the guinea-pig; but the 

 females do. In other words, there is little, if indeed any, evidence of 

 segregation and recombination of factors for size in these crosses. 



It is interesting to note that the F3 hybrids H wild) of both sexes 

 averaged more variable than the guinea-pig. These expressions of 

 average variability were based upon 16 different coefficients of varia- 

 bility. Back of each coefficient of variability there were from 60 to 

 78 variates. If one interprets the data from a purely statistical point 

 of view, then the | wild hybrids were inherently more variable than 

 the parent guinea-pig and the chances are enormous that this difference 

 is not due to random sampling. However, in interpreting biological 

 data, other considerations are of importance. It is shown that all the 

 coefficients of variability in the | wild and the guinea-pig are extremely 

 small. The averages of the guinea-pigs and | wild, in tables 63 and 64, 



