Transition-state theory (TST) provides an important framework for analyzing and explaining the reaction rates of enzymes. TST, however, needs to account for protein dynamic effects and heterogeneities in enzyme catalysis. We have analyzed the reaction rates of β-galactosidase and β-glucuronidase at the single molecule level by using large arrays of femtoliter-sized chambers. Heterogeneities in individual reaction rates yield information on the intrinsic distribution of the free energy of activation (Δ G‡) in an enzyme ensemble. The broader distribution of Δ G‡ in β-galactosidase compared to β-glucuronidase is attributed to β-galactosidase's multiple catalytic functions as a hydrolase and a transglycosylase. Based on the catalytic mechanism of β-galactosidase, we show that transition-state ensembles do not only contribute to enzyme catalysis but can also channel the catalytic pathway to the formation of different products. We conclude that β-galactosidase is an example of natural evolution, where a new catalytic pathway branches off from an established enzyme function. The functional division of work between enzymatic substates explains why the conformational space represented by the enzyme ensemble is larger than the conformational space that can be sampled by any given enzyme molecule during catalysis.