Pulse radiolysis experiments published several years ago (J. Phys. Chem. A, 2002, 106, 2430) raised the possibility that the carbonate radical formed from reaction of center dot OH radicals with either HCO3- or CO32- might actually exist predominantly as a dimer form, for example, center dot(CO3)(2)(3-). In this work we re-examine the data upon which this suggestion was based and find that the original data analysis is flawed. A major omission of the original analysis is the recombination reaction center dot OH + center dot CO3- -> HOOCO2-. Upon reanalysis of the published data for sodium bicarbonate solutions and analysis of new transient absorption data we are able to establish the rate constant for this reaction up to 250 degrees C. The mechanism for the second-order self-recombination of the carbonate radical has never been convincingly demonstrated. From a combination of literature data and new transient absorption experiments in the 1-400 ms regime, we are able to show that the mechanism involves pre-equilibrium formation of a C2O62- dimer, which dissociates to CO2 and peroxymonocarbon ate anion: center dot CO3- + center dot CO3- <-> C2O62- -> CO2 + O2COO2-center dot CO3- reacts with the product peroxymonocarbonate anion, producing a peroxymonocarbonate radical center dot O2COO-, which can also recombine with the carbonate radical: center dot CO3- + CO42- -> CO4- + CO32-center dot CO3- + center dot CO4- -> C2O72-