108 Miss Lee, Miss Lewenz, and Prof. Pearson. [Nov. 3, 



than intelligent girls of 9 years, and non- intelligent boys of 11 years 

 lower heads than the same class of 9 years ! Frankly, we consider that 

 the memoir is a good illustration of how little can be safely argued 

 from meagre data and a defective statistical theory. 



Taking from our school data the auricular height of 2005 boys, and 

 from the growth table based on the same material, reducing them to 

 the age 12 as standard, we find 



(B.) Auricular Height of Head and Intelligence. 



Below 127 mm 



Totals 



Intelligent. 



Slow. 



Totals. 



481 -5 

 415 



584 -0 

 524-5 



1065-5 

 939-5 



896 -5 



1108-5 



2005 



Whence the correlation = 0*0161. 



There is thus less correlation between auricular height and intelli- 

 gence than between either breadth or length and intelligence ; indeed, 

 it is less than the probable error, and no weight can be laid on it what- 

 ever. The discovery of MM. Vaschide and Pelletier that the auricular 

 height of school children is related to their intelligence seems to us 

 quite incorrect for English boys, and unproven owing to defect of 

 material and method even for French children. 



It has been suggested by a sweeping critic, who clings to the high 

 correlation of intelligence and head size, that our school head-measure- 

 ments are of no value. To this we can only reply that in all cases 

 where the measurements have been in the least doubtful the spanner has 

 been returned and the measurements re-made. Further, if the absence 

 of correlation between intelligence and head-measurements be a proof 

 that the head-measurements have been taken badly or the scale of 

 intelligence wrongly applied, how does it happen that high correlation 

 comes out for the head-measurements of brothers, for all three cases, 

 breadth, length, and height, and that its value is quite in keeping 

 with the correlation between the intelligence of brothers 1 The 

 existence of careless measurement or appreciation would have reduced 

 these correlations also to near zero, as well as those on the characters 

 on the same individual. We are forced to conclude that while our data 

 give surprisingly consistent and uniform results for collateral heredity 

 when we deal with upwards of twenty characters,* about half mental 



* Results for seven mental and three physical characters were given in ' Roy. 

 Soc. Proc.,' vol. 69, p. 155. These numbers have been more than doubled since 

 that paper was published. 



