Sorption and desorption behaviour of methane, carbon dioxide, and mixtures of the two gases has been studied on a set of well-characterised coals from the Argonne Premium Coal Programme. The coal samples cover a maturity range from 0.25% to 1.68% vitrinite reflectance. The maceral compositions were dominated by vitrinite (85% to 91%). Inertinite contents ranged from 8% to 11% and liptinite contents around 1% with one exception (Illinois coal, 5%). All sorption experiments were performed on powdered ( -100 mesh), dry coal samples. Single component sorption/desorption measurements were carried out at 22 degreesC up to final pressures around 51 bar (5.1 MPa) for CO2 (subcritical state) and I 10 bar (I I MPa) for methane. The ratios of the final sorption capacities for pure CO2 and methane (in molar units) on the five coal samples vary between 1.15 and 3.16. The lowest ratio (1.15) was found for the North Dakota Beulah-Zap lignite (VRr = 0.25%) and the highest ratios (2.7 and 3.16) were encountered for the low-rank coals (VRr 0.32% and 0.48%) while the ratio decreases to 1.6-1.7 for the highest rank coals in this series. Desorption isotherms for CH4 and CO2 were measured immediately after the corresponding sorption isotherms. They generally lie above the sorption isotherms. The degree of hysteresis, i.e. deviation of sorption and desorption isotherms, varies and shows no dependence on coal rank. Adsorption tests with CH4/CO2 mixtures were conducted to study the degree of preferential sorption of these two gases on coals of different rank. These experiments were performed on dry coals at 45 degreesC and pressures up to 180 bar (18 MPa). For the highest rank samples of this sequence preferential sorption behaviour was "as expected", i.e. preferential adsorption Of CO2 and preferential desorption of CH4 were observed. For the low rank samples, however, preferential adsorption of CH4 was found in the low pressure range and preferential desorption Of CO2 over the entire pressure range. Follow-up tests for single gas CO2. sorption measurements consistently showed a significant increase in sorption capacity for re-runs on the same sample. This phenomenon could be due to extraction of volatile coal components by CO2 in the first experiment. Reproducibility tests with methane and CO2 using fresh sample material in each experiment did not show this effect. (C) 2003 Elsevier B.V. All rights reserved.