No. 604] SHORTER ARTICLES AND DISCUSSIONS 243 



Early Blight, 1905, and Late Blight, 1905. — .066 ± .059 

 Early Blight, 1906, and Late Blight, 1905. . + .190 ± .057 

 Early Blight, 1907, and Late Blight, 1905. . — .040 ± .059 



The results are not so consistent as those of the preceding 

 table. The two negative coeiSeients are insignificant in com- 

 parison with their probahle errors, and the positive one is not 

 large, either absolutely or relatively. Possibly the laxness of 

 the correlation is in part due to the fact that the measurement 

 of both characters is subject to a large possible error.^ 



For freedom of the tubers from scab with the incidence of 

 other diseases every possible correlation has been determined. 

 The coefficients are shown in the accompanying table. Note that 

 in this case the correlation is between freedom from one disease 

 and occurrence of another disease. Hence a negative coefficient 

 has the same meaning as a positive one in the foregoing discus- 



Beeause of the small number of varieties involved and the 

 roughness of the measurements the correlations are low and 

 irregular. In ten cases the negative sign indicates that the 

 varieties which are most free from scab are also least susceptible 

 to attacks by other diseases. In neither of the two eases of posi- 

 tive correlation is the constant statistically significant in com- 

 parison with its probable error. 



Thus altogether 23 of these cross correlal ions— that is correla- 

 tions between injury to different oi-L^nns by tlu- same disease, or 



9 In one case, 1905, the correl.ntion is hctwoi'ii foliairo injury by two dif- 



