||<em>General considerations</em><p/>A selection index may be applied to selection based either upon the performance of single plants or upon the performance of progenies. In the latter case however, index selection should only result in a modest gain in selection efficiency, when index selection is compared to single-trait selection for the desired character.<p/>Index selection is a relatively efficient method of selection when it is applied in the selection of individual plants. Individual index selection, as we may call this type of selection, is presented in this study as a selection method for the initial stage of a population improvement programme. In general it may be used as a method of selection for cross-pollinating crops. Individual index selection combines the advantages of mass selection with these from index selection.<p/>The advantages of mass selection are most obvious during the initial stage of a population improvement programme. Mass selection takes only one generation per cycle. A large number of generations can be grown in a relatively short period of time. When the population under selection is not yet at linkage equilibrium the successive generations will give a high frequency of recombinants implying a progressive breaking up of linkage blocks. With mass selection it is possible, even with limited facilities, to include a large number of plants in the selection, thereby preventing any premature loss of genetic variability.<p/>There are several arguments for the application of a selection index in mass selection. The selection index makes an optimal use of the information available from a single phenotype. With index selection it is possible to select for the simultaneous improvement of several traits. When the selection index is based on traits which may be observed before flowering it is possible to select not only on the seed parent but also on the pollen parent, thereby doubling the response to selection. Furthermore, index selection may be used as a method to select under conditions differing from the normal environment of the crop. This means that it is possible to choose the selection conditions such as to maximize the single-plant heritability and this means then also that it is possible to select in an off-season crop.<p/>Selection indices for individual selection in maize were calculated using estimates of the population parameters obtained either from a parent-offspring relationship or from a sib analysis. There are several advantages in using a parent-offspring relationship. Using a parent-offspring relationship the selection index can be estimated directly from the multiple regression of the value of the offspring on the characters of the parents, independent of any quantitative genetic interpretation of the data. The experimental procedure may be kept rather simple. It is possible to obtain selection indices for the selection under conditions differing from the normal environment of the crop. The latter may be realized by growing the parent generation in the selection environment, while the offspring is measured in the normal cropping environment. <em></em><p/><em>Results</em><p/>The parent-offspring relationship was studied on one hundred and twenty-one open-pollinated plants of Samaru Composite 2 and on thirty-six open-pollinated plants of Colombian Composite. Recorded from each plant were twelve different characters and the mean yield of its progeny. These records were used to construct several selection indices for the improvement of yield. The response to index selection was studied using Colombian Composite as the test population for the selection indices calculated for Samaru Composite 2, and vice versa. The response to individual selection could be improved by between thirty-five and seventy-one percent through the use of a selection index. An optimal response was obtained from selection indices including about four different traits. The response to selection obtained from a selection index was superior to the response to selection obtained from an 'estimated' index, that is an index in which the weighing factors for the different traits represented in the index are estimated according to some rule of thumb.<p/>For the populations under study, composites of recent origin, it was shown that the response to individual index selection compared favourably with the response from selection methods based on progeny means.<p/>A sib analysis was carried out using one hundred and forty-four half-sib progenies of Colombian Composite. Most of the progenies were lost as a result of the sudden outbreak of a disease, caused by <em>Helminthosporium maydis</em> race T, so that the study had to be limited to the results obtained from only thirty-six progenies, resistant to the disease.<p/>The sib analysis was used to estimate the single-plant heritabilities and the genetic and phenotypic correlations with yield and with resistance to stem- and rootlodging for thirty different traits. It was found that the character leaf angle was correlated with yield, which indicates that at Mokwa, despite the rather low yield level of the crop, light is already a limiting factor in maize production. The stalk characters crushing strength and dry weight of a 5-cm stalk section were highly correlated with resistance to stemlodging. A low correlation was found between resistance to stemlodging and the rind thickness of the stalk. The root weight and the pulling strength of the roots were both correlated with resistance to rootlodging, while these two characters were also correlated with yield and with resistance to stemlodging.<p/>The results of the sib analysis were used to construct selection indices for the improvement of yield, for the improvement of resistance to stemlodging, and for the improvement of resistance to rootlodging, while selection indices were also constructed for the simultaneous improvement of yield and resistance to stem- and rootlodging. A selection index for the three simultaneous objectives included first of all the character root weight, secondly it included either the crushing strength of the stalk or the dry weight of a 5-cm stalk section, and thirdly it included the leaf angle. When the selection is aimed at the simultaneous improvement of yield and resistance to stem- and rootlodging, single-trait selection for root weight should be an attractive alternative to individual index selection.<p/>Out of the thirty different traits used in this study there were four traits which could be observed before flowering. These were: the days to shooting, the total number of leaves and the length and width of leaf number eight. Individual index selection before flowering should be efficient in improving yield, when using a selection index based on length and width of leaf eight. As an alternative to index selection one might practise also single trait selection for leaf area, measuring the leaf area as the product of length and width of leaf eight.