The relationship between predicted performance of double crosses of corn in one year with predicted and actual performance of double crosses in later years

1. A comparison was made of the predicted performance of 49 double crosses obtained from the average of four of the six possible single crosses that can be made between four inbred lines not using the parental single crosses in the average and the actual performance of the same double crosses. Predictions were made in both 1940 and in 1941 from yield trials of single crosses grown at four locations, with three replications per location each year. Actual double crosses were tested in 1943 to 1944, inclusive, at five locations with three replications at four locations and two replications at one location. The characters studied included date silked, plant height, ear height ear length, kernel rows, percentage moisture at husking, and yield in bushels per acre. 2. Highly significant correlation coefficients were obtained for predicted performance of 49 double crosses estimated from single crosses grown in 1940 with predicted performance estimated from single crosses grown in 1941. The r values ranged from 0.60 to 0.90 for the characters studied. There were also highly significant r values between the predicted performance in 1940 and 1941, respectively and actual performance in 1943 to 1944, inclusive. 3. The correlation coefficients between predicted performance in one season and predicted performance in another for date silked, plant height, and ear length were significantly higher in one of two comparisons than between predicted and actual performance and in both comparisons for percentage moisture at husking and yield in bushels per acre. There were no differences between the r values for predicted performance in 1940 with 1941 and predicted performance in 1940 or 1941, respectively, with actual performance in 1943 to 1944, inclusive, for ear height or kernel rows. 4. The variability of predicted double crosses was estimated in 1940 and in 1941, respectively, for each of 49 double crosses by comparing the means for the various characters studied of the four single crosses not used as parents of the double cross. There were highly significant relationships between the predicted variability in 1940 and that of 1941. The r values for the characters studied ranged from 0.38 for ear height to 0.92 for percentage moisture. 5. The variability of predicted double crosses in 1940 and 1941, respectively, was compared with the estimated variability of the 49 double crosses grown in 1943 to 1944, inclusive. Significant although not highly important r values were obtained between predicted variability for moisture content in 1940 and in 1941 and estimated variability of the actual double crosses in 1943 to 1944, inclusive, one value 0.39 being significant at the 1% point and the other 0.33 being slightly below the 1% level. One other r value, 0.28, between the predicted variability of double crosses for plant height in 1941 and the actual variability in 1943 to 1944, inclusive, reached the 5% point. Other r values for the relation between predicted variability of double crosses and actual estimated variability for plant height, ear height, date silked, ear length, and kernel rows were of no significance.