Abstract
Demands for service at a public facility during some unit time are assumed to be independent, normally distributed random variables, with a parabolic mean function of time and constant variance. The results indicate that counts for both non-overlapping short time demands are needed to relate fluctuations in demand to capacity. Also, the choice of the length of observation for the average demand is not very consequential for the variance of the observed ratio of the maximum fluctuations to the average demand or the variance of its difference with the appropriate fraction of the average, although the former tends to be smaller for longer observations of the maximum demand.