15. Heterosis of number of kernel rows. - In every one of the crosses of the #1 8-row line with the ten 12-row lines, the average row-number of the two parent lines is greater than that of the corresponding F1. Of the ten F1's involving the same 12-row lines with 9-row line 51, four have mean row-numbers greater than, four less than, and two equal to the average of the two parental lines. It is perhaps noteworthy that the F1 mean of the 1-2 cross differs from the parental average by -1.8 rows, of the 1-39 cross by -0.16, of the 51-2. cross by +0.35, and of the 51-39 cross by +1.38. If the last of these crosses alone had been under observation the rest might well have been termed heterosis - and perhaps correctly so. There is certainly nothing in the general averages to suggest heterosis of row-number. The average of all F1's involving line 1 is less than the average of parental means by 0.81 rows and of those involving line 51 is greater than the parental averages by one 0.17 rows.

Records were also obtained last season from F1 cultures whose parental lines had approximately equal numbers of kernel-rows. The data are given in the accompanying table showing the mean number of kernel rows of inbred lines and their F1 progenies.

 Inbred lines F1 progenies Designation Meannumberrows Average Meannumberrows Differences 1 7.84) 7.90 8.10 0.20 51 7.95) 2 12.04) 12.21 12.41 0.20 4 12.38) 2 12.04) 12.19 12.61 0.42 II 12.34) 2 12.04) 12.03 12.37 0.34 39 12.02) 39 12.02) 12.20 12.58 0.38 4 12.38) 39 12.02) 12.18 13.19 1.01 II 12.34) II 12.34) 12.36 12.53 0.20 4 12.38)

Individually, most of these differences in number of kernel rows are not statistically significant. They are, however, all positive and, as a whole, are definitely significant. In general it appears, therefore, that some, though slight, heterosis is shown in number of kernel rows.