EARLY TESTING AND RECURRENT SELECTION 401 



ear height, and leaf area. These characters, for which selection was efifective, 

 differ from yield in that the genetic basis is undoubtedly much less complex 

 and environmental variability less likely to lead to mistaken classifications. 

 We now know that the plot technics used in these ear-to-row trials were quite 

 inadequate, and some of the failure to achieve improvements in yields must 

 certainly be ascribed to this cause. Many of the modifications of the ear-to- 

 row method of breeding which were introduced to minimize inbreeding prob- 

 ably had an opposite effect, and the rate of inbreeding was actually in- 

 creased. On the basis of data now available, it is impossible to fully assess the 

 relative importance of various causes resulting in the ineffectiveness of this 

 method in increasing yields. 



SELECTION WITHIN AND AMONG INBRED PROGENIES 

 The next method tried, and the one still used most extensively, involved 

 selection wdthin and among inbred lines and the evaluation of the lines re- 

 tained in hybrid combinations. Some of the early work which served as a 

 foundation for this breeding method has been reviewed in other chapters of 

 this book. Extensive breeding programs were established at the various sta- 

 tions in the early 1920's, and a large percentage of the lines now used in the 

 production of commercial hybrids had their origin in this early work. 



In the earlier days of these programs any inbred line which could be main- 

 tained was considered to have potential value. As the work progressed it be- 

 came apparent that inbred lines must meet certain minimum standards of 

 performance as lines in order to merit testing in hybrid combinations. Studies 

 were undertaken by Jenkins (1929) and somewhat later by Hayes and John- 

 son (1939) to determine which, if any, characters of the inbred lines were cor- 

 related with yield in hybrid combinations. In the studies reported by Jenkins 

 correlations were used to measure the relationship between (1) various char- 

 acters of the parental inbreds and the same character in their Fi hybrids, and 

 (2) between characters of the parental inbreds and the means of the same 

 characters for all of their crossbred progeny. The results obtained under 1 

 and 2 were somewhat different. In the first case, none of the characters of the 

 parental inbreds were closely related to the yield of their Fi hybrids. The cor- 

 relations reported ranged from — .10 to -|-.24. The correlations between yield 

 of the parents and yield of their Fi hybrids were .14 and .20. Multiple cor- 

 relations considering various grouping of characters of the inbreds and the 

 yield of their hybrids ranged from .20 to .42. 



In the second series which involved characters of the parental lines and the 

 means of the same characters for all crossbred progeny, the correlations ob- 

 tained were materially larger. With different groups of material the correla- 

 tions involving yield ranged from .25 to .67. In some cases the degrees of free- 

 dom were few and the relationshij) therefore poorly determined. A weighted r 

 calculated for the entire series was .45. The difference between these two 



