172 



Table XII. 



Dr. E. W. A. Walker. 



. — Girls re-grouped — Data and Calculations. 



Group. 



No. in 

 group. 



Average 

 body- 

 weight, 

 W. 



Average 

 body-length 

 (stem-length), 

 I. 



Length 

 constant 



calculated. 



k = Wjl. 



n = -323. 



Body-length 

 calculated. 

 I = 7rW". 

 n = -323. 

 k = 24 -80. 







grm. 



1 



8 



62,960 



2 



29 



52,540 



3 



20 



42,350 



4 



10 



27,250 



5 



12 



15,440 



6 



7 



6,070 



mm. 

 868 -2 

 826 -3 

 774-5 

 692 -2 

 551 -5 

 413 -1 



24 

 24 

 24 

 25 

 24 

 24 



•47 

 •69 

 •80 

 •57 

 •47 

 •78 



mm. 

 880-2 

 830 -3 

 774 -5 

 671 '6 

 559 -0 

 413 -5 



Average 



24 -80 



Average, taking into account number of individuals in 



each group 



Mean deviation (for groups) 



In the latter case it corresponds to a body-weight of 16,220 grm. and a 

 stem-length of 570"4 mm. It is evident that with the data at present in 

 hand it is not practicable to determine the crossing point of the curves more 

 accurately than this. All that can be said at present is that the data for 

 boys and girls show that a crossing takes place somewhere in the region 

 indicated. This may also be seen by comparing the series of body-weights 

 and body-lengths in Table II with those in Table X. Below the point of 

 crossing males are for any given body-length somewhat heavier than females, 

 the difference becoming more marked as the body-length diminishes. Above 

 the crossing the males are for any given body-length somewhat lighter than 

 females, the difference increasing as the body-length increases. 



Distribution of Errors. 



I have examined the distribution of errors in the foregoing series of 

 observations by noting the percentage deviation of the observed body-length 

 of each individual boy and girl from the theoretical value of the body-length 

 as calculated by the formula appropriate to the sex. From considerations 

 of space the calculations are not tabulated here, but the results of this 

 examination are set out in Table XIII as compared with the theoretical 

 distribution of errors when the mean deviation is calculated by the method 

 of least squares. The number of observations in which the percentage 

 deviation falls within - 5, 1, 1*5, 2, 3, and 4 times the mean deviation are 



