In studying synchrony, parameters to vary randomly were chosen in order to have heterogeneity of cell size and growth rate in the culture. For the coefficient of variation of the volume-at-division, a typical value of 0.1 (Koch & Schaechter, 1962; Koch, 1993; Koch, 1996) was used. (Available data give the coefficient of variation (CV) of length-at-division. The CV for volume-at-division is almost as high as the CV for length-at-division if the variability of the width-at-division is assumed to be small compared with length-at-division.) We are not aware of quantitative data for the CV of Vmax that are independent of the variation of size-at-division or which would allow a correction. Hence, we choose a value of 0.1. Jaworska et al. (1996) structured the population similarly by varying the maximal anabolic rate and minimum reproductive volume. Random variation of Vmax led to complete desynchronisation whereas random variation of the volume-at-division did not (Fig. 5). Variation of volume-at-division can be regarded as an imprecision of volume measurement by the cell and broadens the size distribution and more so the age distribution. This view is substantiated since fluctuations in the size at one division are not correlated with the fluctuations in size at the previous division (Schaechter et al., 1962). Size variation does not affect growth rate if uptake and maintenance are proportional to biomass, as holds here. Even when size does not affect growth rate, a correct simulation of average cell size and cell size variation is important since a large cell can survive longer without substrate and divide more readily upon resupply of substrate.