Homepage of project: https://www.uni-potsdam.de/en/emotikon/startseite Children’s development of physical fitness and related moderating effects of age and sex are well documented, especially boys’ and girls’ divergence during puberty. At what age can we detect the onset of this development? In the ninth year of life, sexual hormones start to rise for girls, but not yet for boys. Therefore, this could be the first year in life when girls start to outperform boys, giving rise to a convergence of physical fitness which a few years later will diverge dramatically in favour of boys. Our study had the statistical power to pick up even the weakest signal of this kind. We did not pick up the signal – despite an abundance of statistical power! Along the way, however, we established very strictly linear short-term ontogenetic trends within this single year of life with this cross-sectional sample. The amount of gain varied across the five tests in a theoretically meaningful way related to the development of underlying physiological demands associated with the tests. The same was true for consistent, but also test-specific differences between boys and girls. These results will be of interest to sports science, but also developmental and educational psychology and physiology. The assembly of the sample required aggregation of children over many schools and across nine cohorts. Fortunately, the implementation of the still ongoing EMOTIKON research project with a mandate by the Ministry of Education, Youth and Sport of the Federal State of Brandenburg, Germany, has ensured this for many years. The methodological challenge, however, has been to take into account sources of variance due to differences between children, schools and cohorts. The linear mixed model analysis affords exactly this, but software to estimate models of the complexity we evaluated here has become available only very recently in the Julia programming language. The primary purpose of using a mixed model analysis was to obtain a “clean” test of age- and sex-related effects and their hypothesized interaction for at least some tests after adjustment for school- and cohort-related effects. The model, however, also yielded important results (1) at the school level: it unveiled a correlation between the average physical fitness of schools’ children and the observed developmental gain and (2) at the cohort level: the replication of a decline in cardiorespiratory endurance and an increase in speed components of physical fitness. Thus, we could show within a single analysis that the model recovers important results with respect to each of the three random factors child, school, and cohort. There has been much debate about the value of publishing null results. The main reason for caution is usually related to lack of statistical power. We submit that this study puts an interesting twist on this debate, because many people would argue (wrongly in our opinion) that given the sample size the absence of a significant interaction is more surprising than its presence. This is an opportunity to set the record straight in two respects: First, given sufficient statistical power, a null result can be of high theoretical value, and, second, even in a gigantic sample it is not a given that “everything will be significant.”