Confidence limits of evolutionary synthesis models - III. On time-integrated quantities

Evolutionary synthesis models are a fundamental tool to interpret the properties of observed stellar systems. In order to achieve a meaningful comparison between models and real data, it is necessary to calibrate the models themselves, i.e. to evaluate the dispersion due to the discreteness of star formation as well as the possible model errors. In this paper we show that linear interpolations in the $\log M - \log t_{k}$ plane, that are customary in the evaluation of isochrones in evolutionary synthesis codes, produce unphysical results. We also show that some of the methods used in the calculation of time-integrated quantities (kinetic energy, and total ejected masses of different elements) may produce unrealistic results. We propose alternative solutions to solve both problems. Moreover, we have quantified the expected dispersion of these quantities due to stochastic effects in stellar populations. As a particular result, we show that the dispersion in the $^{14}\mathrm{N}/^{12}\mathrm{C}$ ratio increases with time.