AbstractThe implementation of the theory of diffusion-controlled growth of ferrite plates in plain-carbon steels is critically assessed. It is found that the use of empirically extrapolated diffusion coefficients, phase boundaries, and thermodynamic functions leads to errors in calculations of growth rate. The errors become most important for low transformation temperatures, leading to exaggerated growth rates. Ways of avoiding these difficulties are suggested, and a new analysis of experimental data indicates that the lengthening of Widmanstatten ferrite plates in Fe–C alloys occurs at a rate which is influenced by the diffusion of carbon in the austenite ahead of the interface, assuming that the plates adopt a tip radius consistent with the maximum growth velocity. However, there is a systematic discrepancy between theory and experiment: plate-growth theory seems to underestimate the lengthening rate by some 5μm s−1. This may have something to do with the lath shape of Widmanstatten ferrite, but an analysis using needle–growth theory does not resolve the problem for data obtained at low lengthening rates. In general, plate–growth theory gives a better explanation of experimental data. The growth of bainite sheaves occurs at a rate much faster than expected from carbon diffusion-controlled growth. If the maximum-velocity hypothesis is incorrect (as it is for dendritic solidification), the above-mentioned discrepancies would be larger.MST/192