This paper describes a Monte Carlo study of the linear enhancement factor for the solvent tracer diffusivity in dilute fcc alloys. The model used is the well-known five-frequency model with isolated solute atoms. It is shown that the analytical treatments by Howard and Manning and by Ishioka and Koiwa have major shortcomings in their handling of correlation effects for certain combinations of the atom-vacancy exchange frequencies. Many of the values of the exchange frequency ratios that have been determined from experiment (...) by way of these treatments are probably in some error. (shrink)