W HEN four satisfactory inbred lines of corn are to be used in a double cross, it,is necessary to know how these can be combined to obtain the best possible double cross.Jenkins, 3 in estimating the probable performance of double crosses, concluded that inbred-variety crosses were advantageous in the selection of inbred lines which will combine best in double crosses. Another of the methods of estimation used was to determine the combining ability of four of the six single crosses which can be produced by the crossing of four inbred lines in all possible combinations. Obviously, in the hybridization of two single crosses, the parent single cross combinations are not involved in the yield of the douple cross. It is, therefore, theoretically possible to obtain three different yielding double crosses from four inbred lines by the use of different single cross parents. For example, the expected yield of a double cross (a X b) (c X d) is obtained from the average yield of the four single crosses a X c, a X d, b X c, and b X d. If the yields of the six single cross combinations from four inbred lines are different, the expected yield of the three possible aouble crosses will also be different. Although the inbred variety cross appears to be a desirable method of selecting inbred lines for use in double crosses, single cross data from any four selected lines should therefore be of value in determining which of the three possible double crosses are most productive.The data herein reported were obtained in I935 at University Southeast Branch Experiment Station, Waseca, and at University Farm, St. Paul, Minn. At the Waseca station four inbred lines, cultures II, I4, 374, and 375, which are being used in a double cross and are in convergent improvement studies, were available in the six single and three double cross combinations. Two three-way crosses were also available. Cultures I I and I4 are inbreds of medium early Minn. I3 and cultures 374 and 375 are inbreds of yellow corn and are extremely late in maturity. From previous tests for several years these inbreds were known to have desirable combining ability. The I I crosses and a check hybrid were grown in six replicates of three:row plats. Only three stalk hills in uniform competition were harvested and in no case was a yield figure based on less than ro hills. The mean yields of the crosses and the analysis of variance for the experiment are given in Table r. In the study made at University Farm, four double crosses from Northwestern Dent inbred lines were grown together with the single crosses needed for the prediction of their yields. The crosses were planned so as to determine the yields obtained from two double