Synopsis Soybean seed quality was usually highest from late dates of planting. Exceptions to this general rule were noted when short season varieties planted early matured before the occurrence of hot, dry weather and when full‐season varieties were planted early but matured considerably after the hot, dry weather had ended. An average of visual ratings for several seed characters and a single overall visual seed quality rating were approximately equal to laboratory germination in predicting relative field emergence.

Greater seed yield may be related to a longer reproduction period (RP) in soybean [Glycine max (L.) Merr.]. Unique combinations of genes controlling photoperiod sensitivity may allow selection for longer RP without delaying date of physiological maturity (R7). Three single‐cross populations of F2‐derived soybean lines were made from a diallel crossing of two photoperiod‐insensitive determinate plant introductions, PI317.334B and PI317.336, and an indeterminate cv. Harosoy 63 in an attempt to develop populations in which the correlations of date of R7 with length of the seed‐filling period (SFP) and RP are small. Genetic variability and response to selection of these traits were also studied. Time from beginning of flowering (R1) to beginning of rapid seed filling (RS) of determinate lines averaged 12.6 days and 7.3 days shorter than those of indeterminate lines in two crosses, while SFP, days from R5 to R7, averaged 11.2 and 6.7 days longer for determinate lines than for indeterminate lines. These results are likely due to the difference in estimating stage R5 (seeds 3 mm long in one of the top four plant nodes) between determinate and indeterminate lines rather than biological differences in SFP between determinates and indeterminates. Associations of SFP and RP with R7 in crosses of Harosoy 63 with either plant introduction were weaker than those previously reported for US‐adapted germplasm. Estimates of realized heritabilities of RP ranged from 0.63 to 1.17, even when RP was adjusted for differences in R7. Realized heritability estimates of SFP ranged from 0.40 to 1.02, except for indeterminates where considerably less genetic variance was observed. Reproduction period and SFP of selected F4 populations were lengthened by F3 selection up to 6.4 and 6.1 days, respectively, relative to population means, without significantly changing R7. These results suggest that selections from two of the three crosses should be useful for incorporation of longer SFP and RP into adapted germplasm to enhance yield potential without delaying date of R7.