240 



THE AMERICAN NATURALIST [Vol. LI 



factor, (c), it is obvious that if there be only a very slight aver- 

 age percentage infection the test of disease resistance will not be 

 a very critical one, whereas the average percentage can not be very 

 high indeed unless conditions are so unfavorable that all varieties 

 are affected. The percentage infection of early blight varies enor- 

 mously from year to year. Thus : 



With an incidence of 3 per cent, one year and of 85 per cent, 

 the following season one can, in view of the considerations men- 

 tioned above, hardly expect to obtain smooth values of the corre- 

 lation coefficient. 



It is interesting to compare these results with those for other 

 maladies of the potato. In the same publication Sturrt gives 

 the results of trials for resistance of tubers to scab. Unfortu- 

 nately the experiments of the second year, 1907, included only 

 20 of the 65 varieties from the first year. Calculations may be 

 based on the percentage of tubers which are free or nearly free 

 from scab. This is much lower the second year. 



The probable error is high because of the fewness of the vari- 

 eties retained in the second year's test, but the correlation is of 

 more than medium value and is relatively about 6 times as large 

 as its probable error. Thus susceptibility to scab is probably to 

 a very considerable extent a varietal character. 



The results for tuber rot tests are not available for successive 

 years, but Stuart has given* the percentage of tuber rot in 89 

 varieties grown on sandy loam and on clay loam soil in 1905. 

 For these T find 



22, Tables VI-VII, 



