The model of bacteria population growth in terms of the logistic equation is considered assuming the plate dilution technique of viable counting in the experiment and random errors in data. For example the Fisher test states that the growth equation and the standard description of the random errors are rejected. The extended analysis of errors is presented in terms of an operator which enables to separate three types of random elements: the stochastic growth of dilutions and sampling. On the basis of the triple repetition of observations the variance of dilution errors is estimated.
