Measurement of the effective dielectric permittivity of porous media provides an elegant method of estimating both the porosity and water content of soils, sediments, and rocks. Obtaining accurate estimates of water content can only be achieved if the models relating the measured permittivity to water content work well. One of the difficulties with testing physical models describing two‐ and three‐phase dielectric mixtures (saturated and unsaturated systems) is obtaining an accurate value for the permittivity of the mineral matrix. No direct measurement method has been presented for rocks and soils, so the solid permittivity value is often an estimate of 5, essentially remaining a fitting parameter. We present a measurement method requiring no preassumed theoretical model. The method requires the measurement of the effective permittivity of a granular material immersed in fluids of differing permittivity and interpolation between the range of data points. Measurements of the permittivity of the mineral matrix of a number of materials are presented, including glass beads (7.6), quartz sand (4.7), calcareous seashell fragments (8.9), hematite ore (18.1), tuff (6.0), and a sandy loam soil (5.1). Estimates based on using mixing models to obtain the solid permittivity are demonstrated to underestimate the permittivity of hematite ore by between 0% and 45%, depending on the model and packing. The method is also suitable for measuring the electrical conductivity of granular minerals by changing the electrical conductivity of the bathing solution. Results for the electrical conductivity of hematite ore (0.040 dS m−1) and carborundum (10.9 dS m−1) are presented.