The author suggested some modifications to the additive main effect and multiplicative interaction (AMMI) method to increase its accuracy for measuring stability of genotypes. Using the suggested modifications, four stability levels could be defined. The four stability levels are high, above average, average and below average. The genotype with a high level of stability should have both the first interaction principal component axis (IPCA 1) and the second interaction principal component axis (IPCA 2) equal zero. The level of stability for a genotype with IPCA 1 equal to zero is considered to be above average. The genotype considered to having an average level of stability if its IPCA 2 is equal to zero. Any genotype with IPCA1 and IPCA 2 not equal zero considered to having below average level of stability.

The modified AMMI method was applied to thirty Egyptian genotypes (Gossypium barbadense L.) grown at five locations. The mean squares for environment (E), genotype (G) and GE interaction were highly significant for all traits. The mean square for IPCA1 was highly significant for all traits. The IPCA 2 mean square was significant for boll weight and highly significant for lint yield. The mean square for IPCA 3 was highly significant for lint yield. Four genotypes (F₅ 691/91, F₆ 795/91, F₆ 744/91 and Giza 75) showed a high level of stability for boll weight. For lint yield, four genotypes (M₅ 491/91, F₆ 678/91, F₅ 720/91 and Giza 81) showed a high level of stability. Nine genotypes (M₅ 496/91, M₅ 507/91, F₅ 689/91, F₅ 720/91, F₆ 794/91, F₆ 744/91, F₆ 757/91, F₇ 893/91 and F₈ 901/91) exhibited a high level of stability for lint percentage.