42o 



THE AMERICAN NATURALIST 



[Vol. XLV 



as yield of grain, hectoliter weight, weight of 1,000 kernels, size 

 of kernel, protein content and " mealiness" or softness of grain. 

 Instead of proceeding by the straightforward method of forming 

 a correlation table and deducing therefrom the coefficient of 

 correlation the author follows the laborious, inaccurate and in- 

 conclusive plan of averages. Virtually what is done is to 

 calculate the observed regression line of one character on another. 

 The general result reached, though in no wise critically supported 

 by the evidence presented, is that all of the purely physical char- 

 acters are correlated together to a high degree. The chemical 

 and chemieo-physieal characters protein content and "mealiness 

 are not demonstrably (by the method used) correlated with 

 other characters, though they are mutually definitely correlated. 

 The chief scientific value of the paper is to illustrate in a striking 

 manner how crude and clumsy were pre-Galtonian methods of 

 ntt.-i.-kintr a simple statistical problem. 



Turning now to corn, we have a number of studies of a more 

 or less biometrical character. Apart from the primarily genetic 

 studies on maize of East, Shull, Collins, and Pearl and Surface 

 which are quantitative in character and to some extent- biometric 

 in the treatment of the data, there have appeared recently two 

 special studies on variation and correlation in this plant. The 

 first of these is the paper of Kietz and Smith and the second that 

 of Ewing. The objects of the two papers are apparently some- 

 what dissimilar. Ewing 's is primarily a biological investigation, 

 whereas Rietz and Smith apparently desire primarily to set forth 

 the method of measuring correlation, and incidently to illustrate 

 these principles by means of some corn data which they have on 

 hand. The only general result of particular biological signifi- 

 cance brought out in the work of Rietz and Smith is that the 

 degree of correlation between various ear characters (length, 

 circumference, number of rows, weight) is very markedly influ- 

 enced by environmental conditions surrounding the growing 

 crop. This paper is to be commended for its clear exposition of 

 the method of calculating a correlation coefficient. 



Ewing 's paper contains more matter of general biological 



