The construction and analysis of a balanced, incomplete block experiment that was carried out to compare cotton varieties and lines is discussed. In this experiment v=b=21, r=k = 5, and d = 5. The symbol v represented the number of varieties tested ; b, the number of blocks ; r, replications ; k, varieties per block, and d, number of times any two given varieties occurred in the same block. Since this type of design does not have replications with a complete set of blocks, it is not possible to separate the component between replications of the experimental error. It is, therefore, preferable to use lattice designs for most experiments designed to compare varieties and use the incomplete block designs only when size of blocks is limited by the nature of the material under test. It is not possible to compare the efficiency of balanced incomplete block designs with that of randomized blocks. Judging by the coefficient of variation that was of 16% in the present experiment, it may be said that the estimates of the mean were within the range of those found in randomized block experiments with cotton varieties.