No. 589] 



NOTES AND LITERATUBE 



63 



used as such during the past several years by some of us in prac- 

 tical problems in which we found it impossible to place reliance 

 upon Yule's coefficients and did not feel warranted, because of 

 underlying assumptions, in depending solely upon the classical 

 four-fold method. But it is a measure given in terms utterly in- 

 comprehensible to the ordinary mind, which is quite incapable of 

 thinking in millions or in multiples of millions. 



What Pearson has done with such brilliancy is to furnish a 

 means in mathematical theory and working tables of passing from 

 the incomprehensible scale of pure probability to the familiar and 

 usable and widely comparable scale of correlation. 



As yet it is too soon to be able to state the results of extensive 

 practical application of the new coefficients, but they should have 

 wide usefulness. 



Both Characters Classified by Rank in Series. — In some cases, 

 neither measurements nor classification of the individuals dealt 

 with in categories are given in the data, but merely their position 

 or rank in the series. 



Rank may be numerically expressed, and the suggestion has 

 been made that the correlation of grades or ranks is a quite legiti- 

 mate measure of interdependence in such cases. Pearson 42 has, 

 however, pointed out the very real difficulties encountered in 

 such work. Those who are tempted to use these methods should 

 acquaint themselves with the dangers as pointed out in this 



One Variate Given by Rank in Series, the Other Measured on 

 a Quantitative Scale.— Such cases are not likely to occur with 

 great frequency in biological work. Possible instances are those 

 in. which one wishes to correlate between position in an intensity 



One Variate Given by Multiple or Broad Categories, the Other 

 by Rank in Series.— Practical applications in biology should be 

 rare. For formulae see the paper by Pearson just cited. 



From the foregoing outline it must be clear that of recent 

 years the conception of correlation has been greatly extended and 

 the possibilities of the praetmal usefulness of correlation methods 



42 Pearson, K., "On Further Methods of Determining Correlation," 

 Draper's Co. Res. Mem., Biom. Ser., IV, Dulan and Co., 1907. 



43 Pearson, K., "On an Extension of the Method of Correlation by Grades 

 or Ranks," Biometrika, 10: 416^18, 1914. 



