The probability that an interval contains an element with given characteristics; the limits between which a specified percentage of measurements are expected to lie. A measure of the uncertainty in values. Often taken as the standard deviation (q.v.). Generally calculated by tracing the estimates of the uncertainties in the various measurements that contribute to determining a value, but measurement uncertainties are not always the principal contributors to uncertainty. A rule for calculating uncertainty in electromagnetic applications is to use confidence levels of 95% for data with high signal-to-noise ratios to 68% for noisy data. See Raiche et al. (1985).