The paper reports an investigation of the decomposition of carbon dioxide on an industrial (I.C.I.) methanol synthesis catalyst, which in effect is oxide-supported polycrystalline copper. In the temperature range 173 to 333 K using the technique of reactive frontal chromatography it is shown that (i) the rate of decomposition of carbon dioxide increases on decreasing the dosing temperature, and (ii) the extent of oxidation of copper by the decomposition of carbon dioxide increases on decreasing the dosing temperature. The former result is in agreement with an earlier conclusion that carbon dioxide decomposition on unsupported polycrystalline copper was precursor-state moderated. The "negative" activation energy obtained is in fact the difference in energy between the heat of adsorption of the precursor state and the surface decomposition activation energy of the adsorbed precursor state, which is the same argument as that used by G. Ertl [in "Catalytic Ammonia Synthesis" (J. R. Jennings, Ed.), p. 123, Plenum, New York, 1991] to explain the negative activation energy for the nitrogen sticking probability on potassium-promoted single crystal iron. The inverted temperature dependence of the extent of the oxidation of copper by the decomposition of carbon dioxide is rationalised on the basis that the oxided copper surface reconstructs in an activated way to an unreactive state. The energy barrier to that reconstruction is too high for it to occur at 173 K (the lowest temperature studied) where the extent of surface oxidation (25% of a monolayer) is therefore solely a function of oxygen poisoning of the reactive surface. The CO/CO2 reactive frontal lineshape at 173 K is exactly the same as that of the N-2/N2O lineshape at 333 K; the fewer oxygen coverage at cessation of reaction by oxygen poisoning for the CO/CO2 reaction shows the reaction to be structure sensitive. Infrared spectroscopic data show the reconstruction to involve the loss of the (110) face and the growth of the (211) face while CO desorption data confirm the activated nature of the reconstruction.