Variability is the common thread of biology, and the most variability is usually found between individuals, the issue of sampling and inference is crucial to most of biology. All the other sciences further ‘downstream’ where the variability rules have inherited the methods used initially within biology. The usual and very nearly always the best, way round the problem of inherent variability is randomisation – using chance to even things out. Randomisation cannot reduce the noise, but it can reduce the chance that it biases the results. Whatever the actual situation, in postgraduates’ experiments, it is likely that the outcome variable in the experiment will have some distribution. The whole purpose of controlling experiments by introducing parameters is to reduce the amount of variability within postgraduates’ measurement. The mathematics of the normal distribution has been very well worked out, and much of statistical analysis relies on the normal distribution.