E.H. Haskell took the initiative for an international program of comparison of beta annual dose measurements involving eight thermoluminescence (TL) dating laboratories during 1982 and 1983. For this experiment, 10 samples of bricks were dispatched to the participating laboratories. Firstly, the methods used in the laboratory at Bordeaux are described: x-ray fluorescence, flame emission, plasma emission, neutron activation analysis. Secondly, the results from all the laboratories are discussed. These results are generally in good agreement with a relative standard deviation in the range 3 to 8%. The most significant discrepancies appear for high potassium-containing samples. It is noted that TL dosimetry gives higher values than those obtained from the other methods used. The meaning of these observations is discussed.