Two varieties of cucumbers—Chicago Pickling and Cubit—were selected to test to what extent limiting the number of fruits which mature per plant would affect the yield and quality of the seed produced.

Experiments were conducted to investigate the relationship between the use of a related and an unrelated single cross as a tester parent in evaluating a group of selected F2 plants of maize. Unlike estimates of variability were obtained for per cent stand, ear height grade, per cent moisture in the grain at harvest, and per cent root lodging. In none of the comparisons was the difference large. Correlation coefficients computed for agronomic characters of the related tester crosses and those same characters in the unrelated tester combinations, in general, were low and of little value for prediction. An analysis of the data suggested that the two tester parents did not yield similar measures of combining ability as regards the ranking of the lines. This lack of agreement between the two testers may be attributed largely to differences in specific combining ability. A seriated sample of lines covering the yield range from frequency distributions for the parental and nonparental tester combinations was chosen for further study. These supplementary studies, conducted in 1944 and 1941, gave similar conclusions. This was of particular interest since the two tests were grown under quite different environmental conditions. The results of these investigations further suggested that the related and unrelated tester parent did not rank the lines in similar order. From these data it was not possible to determine which one of the two tester parents was the better for evaluating lines. The correlation coefficients computed for all possible combinations among the tester parents for several agronomic characters indicated differences in their ability to rank the lines which suggested the advisability of using more than one tester. Formulas were used for computing the average gain in combining ability due to the selection of the apparently best instead of a random line from a sample of n lines. The data indicate that the gain in average combining ability beyond the use of 8 to 10 inbred line testers is very slight. This conclusion is in agreement with that of other workers (2) who used single and double crosses involving a comparable number of inbred lines. Using a similar approach for a constant number of lines it was found that the gain in average combining ability beyond the use of three replications was small. Although these results are based on data from single cross trials it would seem logical to assume that they would be applicable to the testing of double crosses. In drawing inferences from these investigations it is well to remember that they are limited to the lines involved in the tests. The results from these investigations have suggested a study to determine the association of inbred lines as testers as compared to their value as testers when in hybrid combinations. For example, if a double-double cross (eight inbred lines) tester gives a comparable estimate to that obtained when using eight lines individually in top-cross combinations the amount of testing would be greatly reduced. Hull (5) suggests that the better yielding inbred line in commercial use at the present are worthless as testers. He postulates that the masking effects of the dominant desirable allels render them ineffective. Data presented here disagree with this hypothesis. The choice of a tester depends upon the use to which the lines under test are to be subjected.

The Science Service building completed, just painted, front view.

The barn at Stavely sub-station before the corrals were built.

The Science Service building completed, just painted, front and side view.

Clearing snow away from the Science Service building

A house at Stavely sub-station.