Resonant Cavity Perturbation (RCP) methods have traditionally required regularly shaped samples, milled or cast into discs, which limits their use in archaeological contexts where specimens are often irregular and non-replaceable. In this study, we introduce a power law correction that eliminates shape dependency, enabled by low-cost PTFE disc replicas using a cuboid chamber. Fourteen ceramic samples from three compositional groups (Heslington, Torksey Fired, Torksey Raw) were measured. The results show that the permittivity correction derived from the Heslington group (α = 0.0126, β = 3.35, R² = 0.86) generalizes more effectively than a fit tailored specifically to the Torksey group, which yields a higher RMSE of 0.185. Loss factor errors are reduced by an average of 86.1% across all fired specimens, while sample volume exhibits no significant influence on measurement error (γ = -0.082, p = 0.82). Validation using CST simulations reveals a mean absolute difference of 0.043 (1.88%) between experimental and simulated ε' values. For ε'', absolute differences remain below 0.015 for all materials, with larger relative discrepancies attributed to small absolute magnitudes. This non-destructive, rapid technique offers a practical pathway for sorting the approximately 30,000 fragments of Ashurbanipal’s Library, reducing the reassembly complexity by an order of magnitude. Further development of this non-destructive technique for the field of Archaeology for other dielectrics such as bone and ivory is ongoing.