This paper describes the statistical analysis of a series of experiments using a simple biological model (wheat germination in vitro), where a large number of wheat seeds were treated with homeopathic potencies of Arsenic trioxide. Some potencies, such as As2O3 40x, 42x and 45x, have repeatedly shown a significant stimulating effect on germination compared to controls, whereas As2O3 35x has a significant inhibiting effect. In some experiments the seeds were stressed before the experiment with a sublethal dose of the same substance. We performed a statistical analysis, both for stressed and non-stressed seed groups, using Poisson distribution as a suitable model for representing the number of non-germinated seeds in a standard experiment with 33 seeds in the same Petri dish. Finally, we have considered the most repeated potencies (30x and 45x), computing the sample odds ratio (OR) and a 95% confidence interval (CI) for the population OR. Our results show significant reproducible effects of some As2O3 decimal potencies, particularly As2O3 45x. In stressed seeds, even decimal potencies of water seem to give significant results compared to control, whereas high dilutions of As2O3 without potentization never show significant effects.