SOCIETY OF THE UNIVERSITY OF ABEEDEEN. 87 



you that the number of individuals in the first group is about 500, in 

 the second group only about 100 ; the results for the first group are 

 therefore likely to be nearer the true magnitudes for the total popula- 

 tion from which it is drawn than the results for the second group. 

 The one set of results has more weight, so to speak, than the other ; 

 how to allow for that, and how to make an accurate comparison 

 between the two sets is the province of the theory of "probable 

 errors," a somewhat advanced part of the Theory of Probabilities, 

 with which I will not trouble you. But I may say that on applying 

 this theory to the figures of Tables VI. and VII. I find that the 

 differences therein shown are not significant, that is, they do not 

 show any very appreciable divergence between the two groups as 

 regards Variability. 



10. I now come to the subject of cam-elation. I have not at- 

 tempted to calculate all the possible coefficients of correlation, but 

 have confined myself to one in each group ; that, however, is an 

 important one, the coefficient of correlation between head length and 

 head breadth. It is as follows : 



Scottish - -3059 + -0275 



Non-Scottish - -2446 + -0658 



Here again I find that the difference between the two groups is 

 insignificant. 



11. The general conclusion which I think the foregoing facts 

 establish is that, as regards the characters which I have selected, the 

 two groups are practically identical. 



IV. THE Two GROUPS COMPARED WITH OTHER CLASSES OF 



THE POPULATION. 



12. I now come to the next part of my paper, in which I propose 

 to compare Aberdeen students with similar classes of the community, 

 and also with classes very different indeed, 



