Résumé
32 F 5 Families derived from two interspecific cotton crosses, three parental lines and two check varieties were evaluated in season 2003 under three planting dates March 24, April 22 and May 20 at the Experimental Farm of the National Research Centre at Shalakan, Kalubia Governorate, Egypt. This was done in an attempt to; select short season cotton genotypes suitable for late cultivation after winter crops [such as bean or wheat] in Southern Delta and Upper Egypt and, to study the association between all studied traits and to identify the superior genotypes using classical selection index. Results obtained indicated that planting dates mean squares were found to be highly significant for all traits studied except position of first fruiting node. Genotypes mean squares were found to be highly significant for all traits evaluated under the three sowing dates and combined data. Combined analysis over three planting dates revealed highly significant genotypes x environment interactions for all traits except position of the first fruiting node. The average number of days to first flower for all populations studied were significantly decreased as planting date was delayed.Correlation coefficients between all possible pairs of traits studied were different in magnitudes from one planting date to another, also, the genotypic correlations were mostly higher in magnitudes than the phenotypic correlations. Path analysis revealed that boll number / plant had the highest direct effect on seed cotton yield followed by boll weight at the three sowing dates. Estimated value of selection index indicated that the ranking of cotton genotypes are not identical in the three sowing dates. It could be concluded that four lines [1,7, 14 and 27] selected in the late planting date may be exploited in breeding early cotton genotypes which tolerate heat stress in squaring, flowering and boll formation periods