Most DNA polymerases incorporate erroneously a nucleotide at a rate of one per ten thousand, mistakes which are usually corrected by a rereading-and-repairing mechanism. Another enzyme, exonuclease, removes an erroneous nucleotide immediately after being synthesized, hydrolyzing its links. It identifies the error by its impossibility of mating correctly with the model, proceeding to remove it while advancing on the DNA strand in the opposite direction to that of the polymerase. Once the error has been eliminated, the polymerase can attempt again to synthesize the nucleotide which, most probably, will mate correctly this time.
Other mistakes which could have escaped to the rereading-and-repairing mechanism can manifest themselves by anomalies in the double-helix structure, resulting from incorrect complementation between the new strand and the model one. A second repairing mechanism copes with the problem, removing the ill-paired bases and replacing them with correct ones. Residual errors, with a frequency of one per billion, might give rise, in the worst-case scenario, to diverse hereditary pathologies, and in the best, to new favorable characters contributing to the betterment of the species or even the eventual emergence of a new one.