Major errors in thermoluminescence (TL) dating are caused by inadequacies in determining the radiation dosimetry. Therefore, an experiment is carried out in which the natural radiation of pottery is simulated. Pieces of a Roman clay pipe are immersed in an U-bearing (approxi-mately 2% U in radioactive equilibrium) solution for various lengths of time (up to 7 months). The growth of fine-grain TL was measured as a function of the immersion period, i.e, of the radiation dose. The content and distribution of U in the pottery was studied with fission tracks. The potential of this technique is that it provides a laboratory simulation of the energy spectrum and spatial distribution of radioactivity. Thus, some correction factors and the radio-active source calibration may be eliminated in TL dating and the TL age accuracy may be improved. The age which was determined with this immersion technique for the Roman pipe is in good agreement with its known archaeological and conventional TL fine-grain and quartz-inclusion ages. -- AATA