In Part I a new theoretical basis was presented for predicting the time-to-ignition in a system that undergoes spontaneous combustion. A lower bound for the time-to-ignition of a self-igniting material was derived in the universally applicable form: τi = ua 2 e1 / ua (1 - e- Δ / ua 2 ) (1) where time to ignition and ambient temperature are expressed in dimensionless terms. Part II presents experimental support for the equation using the self-ignition of cotton fibre. The data illustrates the convergence of three laboratory-scale studies of cotton to the lower bound relationship at higher degrees of supercriticality. This validates the use of the equation for prediction of safe storage times in large scale stockpiles of material. The method can be applied to any self-heating material to give a series of data that form a family of curves in relation to the lower bound equation. The accuracy of time-to-ignition measurements in the laboratory is shown to be dependent on the suppression of self-heating around the periphery of the material. It is demonstrated both mathematically and experimentally that self-heating occurs at the periphery of a sample before any self-heating begins at its centre. Previously reported ignition times that do not account for peripheral effects are unsuitable for prediction purposes and would fall short of the lower bound equation. A nitrogen atmosphere was used as the pre-treatment to suppress the unwanted temperature profile of the cotton as it warmed up to ambient temperature. While nitrogen reduces the effect it does not suppress peripheral self-heating in the laboratory. Furthermore it does not suppress self-ignition either and there are implications here for industrial scale attempts to control spontaneous combustion with inert atmospheres like nitrogen.