This paper examines foundational assumptions underlying gravitational physics, focusing on how gravitational mass is typically inferred from observed motion and then reused as a causal input. It argues that this inferential structure is circular at the generator level and that, as a result, the fitted mass parameter may absorb multiple unmodeled influences—such as historical momentum input, anisotropy, environmental coupling, dissipation, and model-form error—while preserving high numerical precision. The analysis shows that this “mass-absorber” effect is reinforced by two additional assumptions commonly treated as axiomatic: isotropy of gravitational interaction and universal inverse-square distance dependence. Taken together, these assumptions form a closed inferential system in which discrepancies are systematically redistributed into fitted parameters rather than exposed as diagnostic failures. Because the gravitational constant GGG appears throughout physics, this epistemic structure propagates widely. The paper provides a domain-organized catalog of canonical physics formulas containing GGG, spanning Newtonian gravity, general relativity, gravitational waves, astrophysics, cosmology, and Planck-scale constructions, to clarify the breadth of this propagation. The paper does not claim that gravitational predictions are numerically incorrect, nor does it propose a replacement theory. Instead, it distinguishes observational data from interpretive certainty and argues that the magnitude of uncertainty in many gravitationally inferred quantities cannot be internally bounded. The work is intended as a classificatory and methodological clarification, not a rejection of existing tools.