This is a short technical paper in which the author generalizes some limit constructions in probability (almost everywhere convergence of random functions). The classical probability space is generalized to a probability MV-algebra \((M,m)\), where \(M\) is a \(\sigma\)-complete MV-algebra with product and \(m\) is a faithful state, and a classical random function is generalized to an \(n\)-dimensional observable (a map of the Borel subsets of \(R^n\) into \(M\) which partially preserves the structure of random events). Misprints (e.g. a missing parenthesis) are rather numerous but they can be easily corrected.