Inheritance of kernel‑row number.

During the months preceding Dr. Emerson's death he had been engaged in summarizing his work of many years on the inheritance of number of kernel rows in maize. It has been possible, with this material available which included some tabulated data and a partly completed manuscript, to go back through the records and construct a fairly complete picture of the results he had obtained. On the basis of the information collected a paper has been written and is being submitted for publication. The substance of the paper is outlined below.

The original material used consisted of thirteen 12‑rowed, three 10‑rowed and six 8‑rowed inbreds. Even after many generations of inbreeding each line continued to show variability for row number, e.g., among individuals of the 12‑rowed lines there were characteristically many with 10 and with 14 rows, and somewhat fewer with 8 and 16 rows on the ear. The lines were tested for homozygosity of row‑number genes by demonstrating that selection of ears with diverse row numbers as parents had no differential effect on the mean row number of their progeny. Similar tests were conducted on F₁'s between inbred lines with the same result; namely, that selection was ineffective.

The thirteen 12‑rowed genotypes were crossed in all possible combinations to produce 78 F₁'s. Of these 76 had deviations in the plus direction from the mid‑parent value. The average increment was +0.85 rows. The six 8‑rowed genotypes were also crossed in all possible combinations to produce 15 F₁'s. Of these 10 had deviations in the plus direction from the mid‑parent value. The average increment was +0.09 rows. The 8‑ and 12‑rowed genes were then intercrossed in all possible combinations (except one) to produce 77 F₁'s. Of these 45 deviated from the mid‑parent value in a minus direction. The average increment was ‑0.15 rows. The deviations, with the possible exception of the data on 8‑x8‑rowed crosses, were statistically significant.

A plausible interpretation of these results was considered to be: (1)

that the 8‑rowed genotypes contained a preponderance of dominant genes for decreasing row number, that they differed from each other by relatively few of these genes and that the slight plus increment of average F₁ over mid‑parent value could be accounted for by general hybrid vigor of the plant; (2) that the 12‑rowed genotypes contained a more nearly equal number of dominant plus and minus modifiers for row number and that the greater plus increment of average F₁ over mid‑parent value was due primarily to general hybrid vigor (not as greatly counteracted by a preponderance of dominant minus modifiers as in the 8‑rowed lines), and (3) that the average minus deviation of the 8‑x12‑rowed F₁'s from the mid‑parent value could be accounted for by a preponderance of minus modifiers by which the 8- and 12‑rowed genotypes were presumed to differ (sufficient in number or magnitude of effect to more than counteract the influence of hybrid vigor).

In order to determine whether the 12‑rowed phenotypes differed from each other by genes for row number, segregating progenies (usually F₃ to F₅) from 66 F₁'s (all possible combinations among 12 of the 13 inbreds) were selected for high and for low row number. Selection was effective in shifting the mean in at least 63 of these progenies. It was concluded, therefore, that 10 of the inbreds each differed from the other eleven by genes for row number and the remaining two could be the same as only one other line. This being the case, it should have been possible to accumulate additional modifiers in either direction by selection in progeny derived from crosses involving many lines.

Six extracted lines, each derived from crossing four original 12‑rowed inbreds together, were developed by inbreeding and selection for a high number of rows for five generations. The means of these four‑line derivatives were raised above that possible in selections from crosses between two lines. Finally, three extracted lines of multiple genotype origin (each derived from intercrosses involving seven of the 12‑rowed inbreds) were developed by inbreeding and selection for high row number for five generations. In the F₅ generation of these three lines the mode was raised to 22 kernel rows. It has been possible to show, therefore, that by hybridization among 12‑rowed phenotypes, recombination and selection, modifying genes for increasing row number were accumulated so that the expression of this character was raised from 12 to 22 kernel rows on the ear.

H. H. Smith