182 



Kleinere Mitteilungen. 



table V 



At first sight it looks as if no correlation whatever were present. 

 Let us now again very carefully examine and analyse the tables then there 

 is still something more unfolded that we cannot controvert, and which 

 clearly determines that a very true correlation is indeed present, yet that 

 the correlation has been generally obscured by the presence of 

 more correlation diagonals, which must be traced out one by one. 

 Then indeed correlation comes into evidence. It needs no demonstration 

 that mere calculation of a correlation-coefficient in such a table will yield 

 nothing, and that this coefficient will determine nothing concerning 

 correlation. 



Let us turn to tables V and VI, then it must be observed that the 

 direction of the diagonals is not always arbitrary, yet that the diagonal 

 lines are inherent to decided maxima. For that reason we must again 



