<p>Geological hypotheses, informed by textbooks, field observations, or empirical insights, are essential in understanding the subsurface geological structures. The characterization of subsurface targets often hinges on the hypotheses regarding shape and physical properties. This study proposes a Bayesian framework designed to falsify the geological hypotheses using drillholes and geophysical gravity and magnetic data. We test the hypotheses of shape for magmatic Ni-Cu sulfide mineralization depicted by sketches from the literature. Our framework involves a quantitative representation of a geological sketch, followed by Monte Carlo sampling to generate a sequence of models that are aligned with the sketch. These model realizations account for uncertainties inherent in the sketch (i.e., hypothesis). Drillhole data serve as a constraint, ensuring that the model realizations also match with lithologies. Geophysical data are an independent falsification tool, not involved in the modeling process but crucial for hypothesis testing. A specific metric is employed to quantitatively evaluate the falsification and support effective decision-making for shapes. Our study progresses to test hypotheses concerning physical properties, specifically focusing on density contrast and magnetic susceptibility. We evaluate whether these properties are homogeneous or heterogeneous across the survey area. In this paper, a synthetic tutorial example demonstrates the efficacy of our framework, where an incorrect hypothetical sketch is falsified using gravity data. In the real case study, we focus on magmatic Ni-Cu sulfide mineralization at the Crystal Lake Gabbro in Ontario, Canada. The results falsify a blob-shaped hypothesis while confirming a funnel-shaped sketch with heterogeneous density contrast and susceptibility.</p>