448 E. J. WELLHAUSEN 



tion selfed lines obtained from them. In this table it is evident that in those 

 families where the Si lines were poor combiners with a given tester, the S3 

 lines obtained from them on the average also tended to be poor. In those 

 families where the Si lines were good combiners, the S3 lines obtained from 

 them were also good. A correlation coefficient of 0.69 was obtained between 

 the topcross yields of Si lines and the average topcross yields of the S3 lines 

 derived from each. This highly significant correlation coefficient, based on the 

 same 138 pairs whose differences were distributed as shown in Table 27.13, 

 indicates a high degree of relationship between the performance of Si and the 

 average performance of lines obtained from each through subsequent genera- 

 tions of inbreeding. It seems, therefore, that tests for combining ability in the 

 So or Si generation would serve to separate the families that are good com- 

 biners from families that are poor combiners with respect to a given tester in 

 the early stages of the inbreeding program. 



CONCLUSIONS AND SUMMARY 



Although the yield results are based on relatively few years' data, it is evi- 

 dent that the methods used in the improvement of corn in Mexico during the 

 six years that the Mexican program has been under way have given excellent 

 results. In some areas considerable improvement in corn yields was obtained 

 by the wider distribution of certain good native, open-pollinated varieties 

 that had developed in isolated areas through chance hybridization and subse- 

 quent natural selection. In areas where two different adapted varieties were 

 available which expressed a certain degree of heterosis when crossed, the 

 formation of double topcrosses offered a means of rapid improvement over 

 the native varieties. 



It has also been shown that excellent three-way or double cross hybrids 

 can be made from first generation selfed lines. Some of these same double 

 crosses in advanced generations have made good synthetics. This means that 

 those farmers who cannot or are unwilling to plant newly crossed seed every 

 year may still have a 12-20 per cent advantage in yield over their native 

 varieties. Hybrids made from crosses of two synthetics, each consisting of a 

 pooled set of closely related Si lines that combined well with a different 

 pooled set of related Si lines, have shown some promise in the greater sim- 

 plification of hybrid seed production. In this way the maintenance of lines for 

 hybrid seed production can be greatly simplified. 



The use of first generation selfed lines in the early stages of a new breeding 

 program obviously has many advantages. Whether such lines can be main- 

 tained for a reasonable period of time without much change in combining 

 ability remains to be determined. So far through a composite sib method of 

 propagation they have been maintained reasonably "pure." 



Data have been presented which indicate that hybrids made from more 

 homozygous lines might be superior to those made with lines selfed only once. 



