We report on a sensitive and reliable ac technique to measure the Seebeck coefficient S of materials, in particular of high Tc superconductors. The small temperature difference between the ends of the sample allows structure in the S(T) curve to be observed. This technique avoids the creation of a component proportional to dS/dT in the measured signal. Such a component has in the past led to erroneous conclusions regarding the high‐temperature superconductor YBCO. A mathematical analysis shows the origin of this component and how it can be avoided. These predictions are confirmed by experiments on a YBCO single crystal.