摘要：A sample was used of 249 pupils at the age of 7 (+/- 2 months) who were described by 256 variables. For the needs of this paper and for an illustration of target functions, the sample was defined by two variants. In one variant, the sample was defined as the ability for continuous running for 3 minutes (MT3M), and in the other variant it was described as explosiveness, defined as a composite of many variables: long jump from a take-off point, throwing a ball into the distance, and running 20 metres from a standing start. For the stated reasons, the offered model in this paper enables the objectivising of determined groups within a sample defined in any way. The solution is a suitable methodological interaction, in a way that such a solution should be created to respect the target functions (a) and taxonomic determination (b). Such a variant primarily respects the multivariant demands in which the groups are recognised. In this sense, we established a Uditax procedure which taxonomises the test subjects on the basis of the spectral decomposition of the matrix of distances of the subjects themselves in standard space and after determining the number of real taxa by the entropic function. This procedure also identifies the representative subjects of such structures. The obtained taxa may, through any analysis of set redundancy (by regression or canonical procedures), be brought into relation with a differently defined set of criteria-related parameters, i.e. the target function. This rounds up the model in a way that the taxa are grouped against the target functions in accordance with the multi-variant characteristics of individual groups. Using the Uditax procedure, groups were isolated, and, in accordance with normal distribution, the so-called underdeveloped and the athletic were hypothetically positioned on extreme sides, and along with them, the stocky and the underdeveloped, and in the middle the light entities. The discriminative analysis of taxonomic dimensions resulted in all four significant discriminative functions which were easily visualised and simple to interpret.
