242 



THE AMERICAN NATURALIST 



[Vol. LI 



volved in the correlation. The problem is, however, a compli- 

 cated one and much more extensive data are needed for a com- 

 plete analysis. 



A problem of very great biological interest as well as of prac- 

 tical importance is that of the specificity of disease resistance. 

 Concretely : Do varieties differ in their susceptibility to a speeifie 

 disease only, or do they differ merely in susceptibility to disease 

 in general? 



A comprehensive and final answer will require far more data 

 than are available and more stringent statistical analysis than 

 can be illustrated here. Some progress can be made by the 

 method of correlation as follows. 



If susceptibility be purely specific there should be no correla- 

 tion between the incidence of disease x in year (or culture) p 

 and disease y in year (or culture) q, although there should be a 

 correlation between the incidence of disease x or disease y in 

 different years or cultures. If, on the other hand, differences in 

 disease resistances from variety to. variety are determined solely 

 by general weakness or vigor of the stocks, one should expect the 

 correlations between the incidence of different diseases in dif- 

 ferent years or cultures to be (within the limits fixed by the 

 errors of measurement and the probable errors of random sam- 

 pling) as high as those between two series of determinations of 

 incidence of one and the same parasite. 



Consider first the relationship between the percentage of 

 foliage injury by early blight in 1905, 1906, and 1907 and the 

 percentage of tuber rot in 1905. The correlations are : 



Early Blight, 1905 167 ± .057 .256 =t .057 



Early Blight, 1906 211 ± .056 .249 ± .067 



Early Blight, 1907 291 ± .054 .440 ± .058 



"Without exception the correlations are positive in sign. While 

 numerically low, the most of them taken individually may be 

 considered statistically significant in comparison with their prob- 

 able errors. 



Thus it seems clear that the varieties with foliage most injured 

 by early blight are also most subject to tuber rot, just as has been 

 shown to be the case in foliage and tuber infection by late blight. 



For foliage infection by early blight in 1905, 1906, and 1907 

 and foliage injury by late blight in 1905 I find : 



