Studies of heat and mass transfer in packed beds, which disagree substantially in their findings, have nearly all been done with beds of regular particles of uniform size, whereas oil-shale retorting involves particles of diverse irregular shapes and sizes. We, in 349 runs, measured mass-transfer rates from naphthalene particles buried in packed beds by passing through air at room temperature. An exact analogy between convection of heat and mass makes it possible to infer heat-transfer coefficients from measured mass-transfer coefficients and fluid properties. Some beds consisted of spheres, naphthalene and inert, of the same, contrasting or distributed sizes. In some runs, naphthalene spheres were buried in beds of crushed shale, some in narrow screen ranges and others with a wide size range. In others, naphthalene lozenges of different shapes were buried in beds of crushed shale in various bed axis orientations. This technique permits calculation of the mass-transfer coefficient for each active particle in the bed rather than, as in most past studies, for the bed as a whole. The data are analyzed by the traditional correlation of Colburn j(D) vs. Reynolds number and by multiple regression of the mass-transfer coefficient on air rate, sizes of active and inert particles, void fraction, and temperature. Principal findings are : local Reynolds number should be based on the active-particle size, not the average for the whole bed differences between shallow and deep beds are not appreciable; mass transfer is 26% faster for spheres and lozenges buried in shale than in all-sphere beds; orientation of lozenges in shale beds has little or no effect on mass-transfer rate; and for mass or heat transfer in shale beds, log(j .epsilon) = -0.0747 -0.6344logN(Re) +0.0592log(2)N(Re).