570 



Journal of Agricultural Research voi. xix, no. h 



Table XXV. — Forty-one highest-yielding of too Early Six Weeks lines in 1918 "• 



a The 3 1 lines printed in bold-face type also appear in Table XXIV. This leaves 10 of these high-yielding 

 lines with poor vine characteristics. Average of group, 21,350 pounds per acre. , 



CONCLUSIONS 



The data presented, so far as the author is able to judge, do not furnish 

 very strong evidence of the presence of high-yielding lines within the 

 common population of the varieties studied. The real test of the exist- 

 ence of such lines is ability to maintain a high-yielding progeny by 

 indiscriminate mass selection. 



Short performance records are fairly reliable in eliminating lines with 

 low-yielding tendencies, but they are not so reliable as a basis upon which 

 to select plus variations if such really exist. 



If degenerate tendencies exist within certain clonal lines and not in 

 others, short performance records are of little value in eliminating the 

 undesirable lines. 



Degenerate individuals appear with such persistent regularity within 

 line selections as to become a real stumblingblock. If there are no 

 exceptions to this rule, before a hill or tuber line can be increased to a 

 point where it is of real value in commercial potato production it is almost 

 certain to contain degenerate types which soon reduce its yielding power 

 to that of the common population of the variety. 



The data presented will not justify an indorsement of the plan of 

 clonal line selection as a practical method of potato-seed improvement. 

 This does not mean that the hill-selection method of choosing potato 

 seed is without merit. Generally speaking, high-yielding hills selected 

 upon production by weight will produce the following season a high- 

 yielding progeny. It does mean, however, that to be effective hill selec- 

 tion becomes an annual task. I believe there is a more practical 

 method of potato-seed improvement. 



Since certain vine characteristics are so closely correlated with yields, 

 selection based on vine development alone promises to be more reliable 

 than selection based on tuber production either by weight or number, 

 and much more practical. 



