The bimolecular kinetics of addition of pyridine to fulvenone (1), benzofulvenone (2), and dibenzofulvenone (3) in pyridine matrixes at 15 to 70 K have been investigated by FTIR matrix isolation spectroscopy. A detailed kinetic analysis reveals a one-to-one correspondence between the disappearance of ketenes and the appearance of the corresponding ketene-pyridine zwitterions ("ketene ylides") formed upon nucleophilic addition. Perturbed second-order kinetics according to the Kohlrausch model were found to describe the data adequately, whereas pseudo-first-order and strict second-order kinetics did not. These results indicate that the bimolecular reaction examined takes place in a heterogeneous medium where distributions of second-order rate constants are obtained. The heterogeneity index, alpha, obtained from the model varies from 0.35 to 0.56. These findings are consistent with the matrix model proposed by Raff in which the structure of the pyridine matrix is composed of inhomogeneous fast sites where the magnitude of the: true reaction rate for chemical reaction exceeds the diffusion rates of both ketene and pyridine molecules. Arrhenius and Eyring thermokinetic parameters obtained from these data reveal that the: barrier for reaction under these conditions is very small (less than 1 kcal mol(-1)) with a very large negative entropy of activation (ca. - 80 cal K-1 mol(-1)). When compared with analogous solution phase kinetic results from laser flash photolysis experiments, it is concluded that the kinetics observed at these low-temperature conditions are largely that of matrix reorganization, This sets a lower limit for the rates and an upper limit for the activation parameters of the actual chemical reactions. Furthermore. identical magnitudes for E-A, Delta H double dagger, and Delta S double dagger are found even when they are calculated from second-order rate constants based on the incorrect bimolecular model applicable to homogeneous media. It is suggested that the technique of dynamic matrix isolation spectroscopy is not adequate to obtain chemically meaningful activation parameters in bimolecular reactions. Caveats are also pointed out concerning the experimental interpretations of nonlinear Arrhenius plots obtained in cryogenic matrixes, implicit assumptions about the constancy of medium properties in the construction of Arrhenius and Eyring plots, and the calculation of thermokinetic parameters from data fits to linear forms of the Arrhenius and Eyring equations.