The present investigation is dealing with the variances of five long staple Egyptian cotton (Gossypium barbadense L.) genotypes, with respect to yield, its components and fiber properties in an old location (Middle Delta) and a new location (North Delta) during 2004 and 2005 seasons. The final goal is to study the possibility of suggestion a modified analysis of randomized complete block design to replace the use of combined analysis. The five genotypes were cultivars, viz. G.85, G.86 and G.89, the others were hybrids, viz (G.89 x G.86) and (G.89 x Pima S-6). Modified analysis depends on the use of the five genotypes twice, however the number of the replicated genotypes remains the same. The modified analysis gave equal results as the traditional combined. In addition the modified analysis does not need using the Bartlett test .

The goal of the present study was suggesting an evaluation of cotton using the least unit of the latin square design (2 x2). Two Egyptian cotton (Gossypium barbadense, L.) cultivars, viz., G80 and G90 were used. Cultivars were evaluated at three different locations (Beni Souif, Minia and Assuit) in Upper Egypt during 2010 and 2011 seasons except for Assuit in 2010 season. The studied traits were seed and lint yield, boll components (dry weight, seed cotton, lint cotton, seeds weight and number of seeds), indices (harvest, seed, lint and lint percentage) and fiber properties (fiber length, micronaire reading and Pressely index). Five of (2 x2) latin square designs were used in individual locations. The data of five (2x2) latin square design were used together to produce cross over designs. G90 significantly surpassed G80 with respect to seed cotton yield in Beni Souif. In both locations, lint percentage of G80 was greater than the G90. In contrast, harvest index of G90 was greater than G80 since it had the lowest value of dry weight per boll than 80 in both locations. On the other hand, the difference between cultivars with respect to the number of seeds per boll was slight except at Assuit where the differences were significant. The results of fiber properties in both locations revealed that G80 had the longest fiber length compared with G90 due to genetic differences between them and it gave a high micronaire value followed by G90 due to coarseness of fibers. The results of multiple regression revealed that the effects of dry weight per boll, the number of seeds per boll and seed index on seed and lint cotton yield were strong for G80. Results of the present study is important for the regional program to evaluate cotton genotypes.