In exploration for shale gas, experimental methane-sorption measurements represent a valuable source of information for resource estimates and for reservoir-modeling studies. Here, the main difficulty is the relatively low adsorption capacity of shales (typically 10% of the sorption capacity of coals), as well as the fact that the measurements need to be performed over a wide range of pressures and temperatures characteristic of past or present geological conditions. In this work, we demonstrate the capabilities of an adapted manometric apparatus to reliably measure excess sorption isotherms at pressures of up to 30 MPa and temperatures up to 423 K on carbonaceous shales. This is accomplished with an experimental design comprising separate heating zones for the sample cell and for the rest of the apparatus. An experimental and mass-balance approach is presented to quantify the temperature gradient existing between the two heating zones, as well as the thermal expansion of the sample cell, and to account for these in the calculation of the excess sorption. We demonstrate that the analysis of the helium-void-volume data over a large temperature range can be interpreted with respect to the thermal expansion of the sample and, in some cases, changes in pore-volume accessibility to helium. We propose to perform blank-expansion tests with non-adsorbing specimens (e.g., steel cylinders) as a quality check to eliminate device-specific artifacts resulting from unknown measurement uncertainties or from uncertainty in the equation of state. Two evaluation procedures are presented to quantitatively account for the blank tests in the final result of sorption measurements on shale samples. As an example, methane-sorption isotherms for carbonaceous shale at 311, 338, 373, and 423 K are presented. By use of a Monte Carlo algorithm to simulate the propagation of the experimental uncertainties, the final estimated uncertainty in excess sorption resulting from systematic errors was found to be 6 0.007 mmol/g at 25 MPa. The consideration of the blank-expansion tests in the mass balance further reduces the systematic error, at least to a point at which an excellent intralaboratory consistency is obtained. The estimated uncertainty resulting from random errors was found to significantly overestimate the actual precision of the experimental setup, and an explanation is provided with respect to experimental design. A datareduction approach using an excess-sorption function based on a Langmuir-type absolute-sorption model was found to provide an excellent representation of the measured sorption data. By means of simplified model calculations we demonstrate that the excess-sorption formalism is a sufficient, simple, and adequate approach to applications in shale-gas-resource estimation. The uncertainties pertaining to representativeness of experimental sorption data of in-situ reservoir conditions are briefly discussed.