Pharmacognosy frequently moves across materials and levels of inference: a botanical extract produces a phenotype, chemical analysis detects candidate constituents, an isolated compound shows activity, and computational or experimental work nominates a molecular target. These observations are scientifically valuable but do not represent equivalent claims. Mechanistic overstatement arises when evidence supporting one level is silently transferred to another—for example, when extract activity is attributed to a detected constituent, a compound–activity association is treated as causal, or biochemical inhibition is described as proof that a target mediates a cellular phenotype. This article develops an original evidence-translation model for keeping these claims analytically separate while still allowing evidence to accumulate across levels. The model treats material definition, chemical identity, constituent attribution, compound activity, target-specific activity, target engagement, and causal mechanistic interpretation as related but non-interchangeable evidential states. Translation between states is proposed to require claim-matched bridging evidence rather than rhetorical continuity. The framework also treats assay interference, mixture interactions, analytical uncertainty, exposure, off-target behavior, and biological context as active constraints on interpretation rather than secondary caveats. Its purpose is not to impose a single experimental sequence or universal proof threshold, but to make the evidential basis of mechanistic language more explicit. Applied consistently, the model can support clearer pharmacognostic reporting and better identification of where additional validation is required. Its boundaries remain important: the framework is conceptual, context-dependent, and not prospectively validated as a scoring or decision system.
