Forty-five lines were crossed to six testers to evaluate their general combining ability. These lines did not clearly represent a random selection of lines with respect to this character, though they had not undergone any formal selection in this respect. In these lines, the variance in specific combining ability exceeded the variance in general combining ability. Six or more testers are required to evaluate general combining ability in lines of this nature. The interaction of general combining ability and locations was less than the variation in SCA and locations.