14 ETHEL M. ELDERTON 



For / there are such changes for both generations ; and yet / is larger than B and D. 

 We must thus consider that a difference of sex in the same generation makes no 

 difference in the intensity of resemblance so far as our present data go. Accordingly, 

 if change of sex in descent does to some extent weaken inheritance*, it does not 

 appear connected with sex differences in the same generation. The differences noted 

 are, however, too slender and the whole system of values too fluctuating to build up 

 any hypothesis as to sex influence in heredity. 



If we now turn to the separate characters, and compare irrespective of cousin type 

 the general means of each, we find an even wider range of results ('19 to '34). We 

 attribute this only in part to real differences in the intensity of resemblance ; we 

 consider it more due to (a) difficulties of estimating some of the characters dealt 

 with, especially as in the case of cousins they are usually not in daily contact with 

 each other; (b) differences of method employed, and the assumption that temperament 

 follows a normal distribution of frequency. 



Accordingly we shall draw no conclusions as to divergences in resemblance, 

 believing our data may be relied upon to give a " general average resemblance " of 

 cousins, but cannot be pressed beyond such a result to discriminate between individual 

 of character. 



6. We now turn to the results of the second series of quantitative measure- 

 ments. These measurements as we have already noted are far from complete. They 

 give for the four measurements 107 pairs of female cousins, 34 pairs of male cousins, 

 and 111 pairs of male and female cousins, the two first sets giving 214 and 68 pairs in 

 the symmetrical table. 



The following table gives the statistical constants of the series of measurements. 

 It will be observed that we have two series for each sex, but not all the individuals in 

 each series are different. 



This table shows at once considerable irregularities, which may be due to the 

 paucity of data, or to the defective handling of the spanners. While the finger joint 

 measurements give a sex-ratio for the absolute lengths = '91, very nearly the usual 

 11/12 of stature and of bone measurements in man, the widths of hand and wrist 

 (involving a good deal more care in determination and allowing of more personal 

 equation) give ratios of about H'5/12. For these also the man is both absolutely and 

 relatively more variable than the woman. For the joint measurements the woman is 

 equally variable absolutely with the man, and relatively more variable. It seems 

 improbable that this equal absolute variability is correct. It is not true for the 

 majority of bone measurements in man and woman. It is further to be noted that in 

 the Male and Female Cousins series, where there was a much larger return of measure- 



* The influence of change of sex has been very elusive ; it would appear to have some bearing on the 

 inheritance of eye-colour in man (Biometrika, Vol. II. pp. 237-40), but we have failed to find it in coat-colour 

 in horses (Ibid. Vol. n. pp. 229-34). It is doubtfully significant in the cases of coat-colour of Greyhounds 

 (Ibid. Vol. in. pp. 257-8), and of Shorthorns (Ibid. Vol. iv. pp. 449-51). 



