Background: Reproducibility of basic research investigations in homeopathy is challenging. This study investigated if formerly observed effects of homeopathically potentised gibberellic acid (GA3) on growth of duckweed (Lemna gibba L.) were reproducible.Methods: Duckweed was grown in potencies (14x–30x) of GA3 and one time succussed and unsuccussed water controls. Outcome parameter area-related growth rate was determined by a computerised image analysis system. Three series including five independent blinded and randomised potency experiments (PE) each were carried out. System stability was controlled by three series of five systematic negative control (SNC) experiments. Gibbosity (a specific growth state of L. gibba) was investigated as possibly essential factor for reactivity of L. gibba towards potentised GA3 in one series of potency and SNC experiments, respectively.Results: Only in the third series with gibbous L. gibba L. we observed a significant effect (p = 0.009, F-test) of the homeopathic treatment. However, growth rate increased in contrast to the former study, and most biologically active potency levels differed. Variability in PE was lower than in SNC experiments. The stability of the experimental system was verified by the SNC experiments.Conclusions: Gibbosity seems to be a necessary condition for reactivity of L. gibba to potentised GA3. Further still unknown conditions seem to govern effect direction and the pattern of active and inactive potency levels. When designing new reproducibility studies, the physiological state of the test organism must be considered. Variability might be an interesting parameter to investigate effects of homeopathic remedies in basic research.